Effectiveness of digital health interventions for the management of Low Anterior Resection Syndrome (LARS) after sphincter-preserving surgery for rectal cancer: a systematic review and meta-analysis

Abstract

Background

Low anterior resection syndrome (LARS) frequently occurs following sphincter-preserving rectal cancer surgery. Although recent studies have reported the effects of digital health interventions for LARS management, no systematic review and meta-analysis has been conducted.

Methods

We searched five English databases and three Chinese databases from inception to 18 December 2025, supplemented by gray literature sources and reference checking. Randomized and non-randomized controlled studies involving adults who had undergone sphincter-preserving rectal surgery were eligible. Risk of bias was assessed using RoB 2 for randomized trials and ROBINS-I for non-randomized studies. When at least two studies reported similar outcomes, meta-analysis was performed using mean difference or standardized mean difference with 95% confidence intervals, with the I² statistic used to measure heterogeneity. Narrative synthesis was used when pooling was not appropriate. Certainty of evidence was assessed using GRADE. The protocol was registered in PROSPERO.

Results

Eight studies published between 2021 and 2025 were included, and seven contributed to the meta-analysis. Digital health interventions were associated with reductions in LARS scores compared with usual care (MD = − 3.00, 95% CI − 4.17 to − 1.83; P < 0.00001; I² = 28%) and with improvements in health-related quality of life, although heterogeneity for this outcome was substantial (SMD = 0.75, 95% CI 0.17 to 1.34; P = 0.01; I² = 89%). No statistically significant between-group effect was observed for self-efficacy, with substantial heterogeneity (SMD = 0.43, 95% CI − 0.15 to 1.01; P = 0.15; I² = 79%). Anxiety and depression outcomes, assessed in a small number of trials, did not show meaningful between-group differences. Evidence for self-management, activation, satisfaction, feasibility, and other implementation outcomes was suggestive but inconsistently reported and often study-specific.

Conclusions

Digital health interventions show promising potential to reduce LARS symptom burden and may improve quality of life after sphincter-preserving rectal surgery. Nonetheless, confidence in these findings is limited by the small number of available trials, clinical and methodological heterogeneity, and inconsistent outcome measurement and reporting, particularly for psychological and intermediate pathway outcomes. Further well-designed, adequately powered studies are needed to confirm effectiveness and inform implementation in routine care.

Trial registration

PROSPERO Registration: CRD420251250663.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12876-026-05258-7.

Background

Sphincter-preserving rectal surgery is increasingly performed as oncologic outcomes improve and surgical techniques advance [1]. Nevertheless, many survivors experience persistent bowel dysfunction, collectively termed low anterior resection syndrome (LARS) [2]. LARS is typically characterized by bowel urgency, stool clustering, increased stool frequency, and flatus or fecal incontinence, and it can substantially compromise daily functioning, social participation, and overall quality of life [3]. Contemporary evidence indicates that bowel dysfunction is highly prevalent after sphincter-preserving rectal cancer surgery, with LARS affecting approximately 60%–90% of patients and major LARS occurring in around 40% in pooled estimates [4]. In a large population-based cohort, the overall prevalence of LARS was 77.4%, and the prevalence of major LARS was 53.1%, underscoring the clinical importance of effective long-term management strategies [5].

Current guidelines for LARS management generally recommend a systematic approach that begins with conservative supportive and behavioral interventions, including dietary adjustments and medication optimization [6]. When symptoms persist, pelvic floor rehabilitation and transanal irrigation are often considered as next-step options, whereas neuromodulation and other advanced therapies are typically reserved for patients with refractory symptoms [7]. Despite this structured escalation approach, LARS remains difficult to manage in routine care. This challenge is driven by symptom variability over time, delays in help-seeking related to embarrassment or stigma, and constrained follow-up capacity, particularly outside specialized centers [8, 9].

Digital health interventions delivered through mobile applications, web-based platforms, instant messaging tools, and telehealth models may help address these care gaps by extending structured education, facilitating repeated symptom assessment, and enabling timely feedback and escalation between clinic visits [10]. Early LARS-specific digital follow-up tools have been described, and more recent trials have evaluated mobile health–based remote interaction management strategies aimed at improving quality of life and self-management capacity [11–13]. However, the existing evidence is fragmented, with substantial heterogeneity in study designs, intervention components and intensity, outcome measures, and follow-up schedules. In addition, implementation-related outcomes are reported inconsistently, which limits the ability to translate current findings into practice-ready recommendations.

Therefore, this systematic review and meta-analysis aimed to synthesize the available evidence on digital health interventions for LARS management after sphincter-preserving rectal surgery. We evaluated their effects on LARS symptom outcomes and patient-centered domains, and summarized implementation-relevant evidence to inform future intervention development and clinical application.

Methods

Design

This review followed the updated Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) statement [14] to report the findings. The protocol was registered on the International Prospective Register of Systematic Reviews (PROSPERO; ID: CRD420251250663).

Search strategy

We conducted a systematic search across five English databases (PubMed, Embase, Web of Science, Cochrane Library, and CINAHL) and three Chinese databases (CNKI, Wanfang Data, and SinoMed) from inception until December 18, 2025. The search used Medical Subject Headings (MeSH) and free-text terms related to “digital health” and “LARS”. In addition, we searched the following gray literature sources: ProQuest Dissertations & Theses Global, ClinicalTrials.gov, and the Chinese Clinical Trial Registry (ChiCTR). The reference lists of all included studies and pertinent reviews were reviewed for additional references. A detailed description of the complete search strategy is provided in the Supplementary material (Supplementary Table 2).

Study eligibility criteria

The predetermined eligibility criteria were defined using the PICOS framework [15]: (1) Population: Adults (≥ 18 years) who had undergone sphincter-preserving rectal surgery for rectal cancer and were experiencing LARS or were at risk of LARS. (2) Intervention: Digital health interventions refer to structured programs delivered primarily via digital or telecommunication technologies, with support from healthcare professionals. Delivery modalities may include mobile applications or mini-programs, web-based platforms, text messaging or email, instant messaging tools, and telehealth follow-up by video or telephone. These interventions typically combine interactive health education with ongoing symptom assessment and feedback, behavioral support, and follow-up management to facilitate sustained management and improvement of LARS symptoms after rectal cancer surgery. In this review, interventions were considered eligible when the digital or telecommunication component was used to support LARS-related clinical management, self-management guidance, symptom monitoring, rehabilitation support, or postoperative follow-up, rather than solely for recruitment, administrative communication, or outcome data collection. (3) Comparison: usual care. (4) Outcomes: the primary outcome was the LARS score. Secondary outcomes included other symptoms and patient-relevant outcomes, such as health-related quality of life, self-efficacy, and psychological outcomes. Additional outcomes, such as satisfaction and feasibility, were included if reported by the included studies. (5) Study design: randomized controlled trials and non-randomized controlled studies were eligible. Only full-text reports available in English or Chinese were included. Study protocols, editorials, commentaries, reviews, conference abstracts, case reports, and qualitative studies were excluded.

Study selection and data extraction

The search results identified from the databases were imported into EndNote X21 software for data management. After duplicates were removed, two reviewers (ZYY and CXY) independently screened the titles and abstracts of all retrieved records according to the pre-established inclusion and exclusion criteria. Disagreements were resolved through discussion, with consultation with an additional review author (NL) when necessary. Full-text articles were then obtained and independently assessed, and only studies meeting all eligibility criteria were included in the final review.

Data extraction was carried out independently by two reviewers (ZYY and CXY) using a standardized Microsoft Excel extraction form, with discrepancies addressed through discussion or consultation with another review author (NL). The extracted information included the first author, publication year, country, participants, study design, sample size, mean age, details of interventions (including intervention content, frequency, and duration), control interventions, assessment time points, outcomes, measurement instruments, and relevant numerical outcome data. In cases where data were missing or incomplete, the corresponding authors of the original studies were contacted by email for clarification.

Quality assessment

Two reviewers (ZYY and CXY) independently assessed the risk of bias in the included studies using the Cochrane Risk of Bias tool (RoB 2) [16] for randomized controlled trials and the Risk Of Bias In Non-randomized Studies of Interventions tool (ROBINS-I) [17] for non-randomized intervention studies. Disagreements were resolved through discussion, and another review author (NL) was consulted when necessary to reach a final decision. The RoB 2 assesses the following five bias domains: the randomization process, deviations from the intended interventions, missing outcome data, measurement, and selective reporting of results. The ROBINS-I assesses six domains, including confounding, intervention classification, participant selection, missing data, outcome measurement, and reporting bias.

Data synthesis and analysis

A meta-analysis was conducted when at least two studies reported the same outcome with available data, using Review Manager (RevMan 5.4). For continuous outcomes, means and standard deviations were extracted whenever available. When medians with interquartile ranges were reported, means and standard deviations were estimated using validated conversion methods [18]. When estimated marginal means with 95% confidence intervals were reported, standard deviations were calculated from the confidence interval width, sample size, and the corresponding t distribution [19]. When multiple follow-up time points were available, data from the latest assessment were extracted. The pooled effect estimates were calculated as mean difference (MD) or standardized mean difference (SMD), with 95% confidence intervals (CIs), based on the similarity of outcome measurements.

Heterogeneity was assessed using Cochran’s Q test and the I² statistic, with the significance threshold set at P < 0.10; I² values of < 50% were considered to indicate low heterogeneity, and I² values of ≥ 50% were considered to indicate substantial heterogeneity. Given the anticipated clinical and methodological heterogeneity across digitally supported LARS management interventions, random-effects models were used for all meta-analyses. Subgroup analyses were conducted to explore potential sources of heterogeneity according to delivery modality, intervention content, and follow-up period. Short-term follow-up was defined as up to 3 months after baseline, and medium/long-term follow-up was defined as more than 3 months after baseline [20]. Additionally, sensitivity analysis was conducted by sequentially removing individual studies to assess the stability of the pooled results. Narrative synthesis was conducted when quantitative pooling was not appropriate because only one study reported the outcome, or because of differences in study design, outcome measures, or substantial heterogeneity.

The standardized data extraction form, the outcome-by-study coverage matrix, and detailed conversion procedures are provided in the Supplementary material (Supplementary Tables 3 and 4).

Assessment of publication bias

Funnel plots and statistical tests for funnel plot asymmetry, including Egger’s and Begg’s tests, were planned to assess potential publication bias when at least 10 studies were available for a given outcome [21, 22]. Because each quantitative synthesis included fewer studies than this threshold, these assessments may have low statistical power and yield unreliable results. However, given the potential vulnerability of digital health intervention studies to publication bias, funnel plot inspection and Egger’s regression test were conducted as exploratory supplementary analyses, and the findings were interpreted cautiously.

Certainty of evidence

The certainty of evidence was evaluated using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach with GRADEpro [23]. Evidence from RCTs was initially rated as high certainty, whereas evidence from non-randomized studies was initially rated as low certainty. The certainty was downgraded by one level for serious (or two levels for very serious) limitations, including risk of bias, inconsistency, indirectness, imprecision, and publication bias.

Results

Search results

As shown in Fig. 1, a total of 1359 articles were identified in the initial search. After removing duplicates, 803 articles remained for title and abstract screening. A full-text review was conducted on 40 articles, of which 8 studies met the eligibility criteria and were included in this review. Of these, 7 studies were deemed suitable for meta-analysis.

Fig. 1.

PRISMA 2020 flow diagram

Characteristics of the included studies

The characteristics of the included studies are summarized in the Supplementary material (Supplementary Table 1). The 8 studies were published between 2021 and 2025. The studies were conducted in China [11, 12, 24], the United States [25, 26], Canada [13, 27], and the Republic of Korea [28]. The evidence base comprised five randomized controlled trials [12, 13, 25, 27, 28], two quasi-experimental studies [24, 26], and one observational cohort study [11]. Participants were adults who had undergone sphincter-preserving surgery for rectal cancer and were experiencing LARS (or related postoperative bowel dysfunction). The sample sizes ranged from 15 to 160, with mean ages ranging from 54.6 to 65.2 years.

Intervention characteristics

The included interventions were delivered through different digital or remote approaches, mainly including mobile apps [12], online platforms [13], WeChat-based programs [11, 24], telephone-based support with SMS/email follow-up [25, 27, 28], and telehealth coaching [26]. The main LARS management components included LARS-related education [12, 13, 24, 25, 27, 28], bowel symptom assessment, diary recording, or self-monitoring [11–13, 25, 27, 28], dietary or lifestyle guidance [12, 24–26, 28], pelvic floor or bowel rehabilitation [24, 28], professional or nurse-led support [12, 24, 27, 28], peer support [13, 24], and reminders or remote follow-up [11, 12, 24, 25, 27, 28]. Because most interventions were multicomponent and included overlapping elements, the core LARS management components of each study were mapped in Supplementary Table 8. Follow-up time points varied across studies, ranging from discharge to 12 months. Detailed intervention characteristics are provided in Supplementary Table 1.

Each intervention was assigned to one dominant functional category according to its main LARS management focus, intervention aim, and therapeutic content. Three dominant functional categories were identified: supportive self-management and remote follow-up [11–13, 27], lifestyle behavior–focused symptom management [25, 26], and bowel rehabilitation–focused multicomponent care [24, 28]. Interventions were grouped into three delivery modes: app- or web-based platforms [12, 13], social media-based tools [11, 24], and telecommunication-based support [25–28]. The classification of interventions is presented in Table 1.

Table 1.

Classification of included interventions according to mode of delivery and LARS management function

Study	Mode of delivery	Dominant LARS management function
Ruan et al. 2025	Social media-based tools	Bowel rehabilitation–focused multicomponent care
Garfinkle et al. 2025	Telecommunication-based support	Supportive self-management and remote follow-up
Moon et al. 2025	App- or web-based platforms	Supportive self-management and remote follow-up
Nehemiah et al. 2025	Telecommunication-based support	Lifestyle behavior–focused symptom management
Zhou et al. 2024	App- or web-based platforms	Supportive self-management and remote follow-up
Sun et al. 2024	Telecommunication-based support	Lifestyle behavior–focused symptom management
Kim et al. 2023	Telecommunication-based support	Bowel rehabilitation–focused multicomponent care
Liu et al. 2021	Social media-based tools	Supportive self-management and remote follow-up

Risk of bias in the included studies

Among the five randomized controlled trials, two were rated as high risk [13, 28], two as some concerns [12, 25], and one as low risk of bias [27] (Fig. 2A). The trial by Kim et al. [28] was rated as high risk primarily because allocation concealment was not ensured, and outcome measurement was likely influenced by the lack of blinded assessment. The study by Moon et al. [13] was judged to be at high risk due to concerns related to missing outcome data, coupled with a high risk of bias in outcome measurement and selective reporting. Zhou et al. [12] and Sun et al. [25] were considered to present some concerns, mainly due to limited information on prespecified analysis plans and uncertainty regarding the selection of reported results.

Fig. 2.

Risk of bias assessments for each included study: (A) randomized controlled trials; (B) non-randomized controlled trials

For the three non-randomized studies assessed using ROBINS-I, one was judged to be at overall moderate risk of bias [24], whereas two were judged to be at overall critical risk of bias [11, 26] (Fig. 2B). The moderate-risk study was mainly limited by potential bias in outcome measurement, while key domains such as confounding and missing data were judged less problematic [24]. In contrast, Nehemiah et al.’s [26] study was rated as being at critical risk of bias primarily because of uncontrolled confounding, with further concerns related to participant selection and missing data. The study by Liu et al. [11] was judged to be at critical risk largely because of serious risks arising from missing outcome data and selective reporting, with further concerns about selection into the analysis. More details of the risk of bias assessments are presented in the Supplementary material (Supplementary Tables 5 and 6).

Effectiveness of digital health interventions

Primary outcome: LARS score

All included studies reported the LARS score [11–13, 24–28]. However, only six were eligible for meta-analysis. The study by Nehemiah et al. [26] was excluded because it used a single-arm pre-post design without a concurrent control group, precluding estimation of a between-group effect. Garfinkle et al. [27] reported dichotomous LARS outcomes. The authors were contacted by email to request continuous LARS score data, but no response was received. Consequently, quantitative synthesis was restricted to six controlled intervention studies with comparable continuous LARS score data, whereas Garfinkle et al. [27] and Nehemiah et al. [26] were summarized narratively. The pooled analysis indicated that digital health interventions were associated with an improvement in LARS symptom burden (MD = − 3.00, 95% CI − 4.17 to − 1.83; P < 0.00001; I² = 28%) (Fig. 3). The certainty of evidence was rated as moderate, primarily because several trials showed methodological limitations and the total information size remained limited, leading to concerns regarding risk of bias and imprecision (Supplementary Table 7).

Fig. 3.

Forest plots: effect of digital health interventions on LARS score

Subgroup analyses showed no statistically significant subgroup differences according to delivery modality (Chi² = 1.90, df = 2; P = 0.39; I² = 0%) (Supplementary Fig. 1), dominant LARS management function (Chi² = 1.39, df = 2; P = 0.50; I² = 0%) (Supplementary Fig. 2), or follow-up duration (Chi² = 0.66, df = 1; P = 0.42; I² = 0%) (Supplementary Fig. 3). These findings provided no evidence that the intervention effect on LARS score differed across the exploratory subgroups. Given the limited number of studies within each subgroup, these results should be interpreted cautiously. Leave-one-out sensitivity analysis showed that the pooled effect for LARS score remained statistically significant after sequential exclusion of each study, with MDs ranging from − 2.83 to − 3.31. I² values ranged from 0% to 42%. The exclusion of Moon et al. reduced heterogeneity from 28% to 0% and yielded a similar pooled effect (MD = − 3.31, 95% CI − 4.13 to − 2.48), suggesting that this study may have been the main contributor to heterogeneity. Overall, the findings were robust (Supplementary Table 9). Consistent with the overall direction of effect, the two studies not pooled also suggested potential benefit, albeit reported in different formats. Specifically, in Garfinkle et al. [27], the incidence of major LARS was lower in the intervention group at 1 month. Similarly, Nehemiah et al. [26] reported a significant reduction in LARS from baseline at 12 weeks, and the improvement was maintained at 24 weeks.

Secondary outcomes

Health-related quality of life

Seven studies assessed health-related quality of life using three instruments: four used the EORTC QLQ-C30 [12, 13, 24, 27], two used the COH-QOL-CRC [25, 26], and one used the EORTC QLQ-CR29 [28]. Three studies were excluded from the HRQoL meta-analysis and summarized narratively. Ruan et al. [24] did not report a global HRQoL score; Kim et al. [28] used the EORTC QLQ-CR29, which does not provide a standard global or total HRQoL score; and Nehemiah et al. [26] was a single-arm study without a control group. Four studies reported comparable global or overall HRQoL data and were included in the meta-analysis. The random-effects model showed that digital health interventions were associated with significantly higher HRQoL scores compared with usual care, although heterogeneity was substantial (SMD = 0.75, 95% CI 0.17 to 1.34; P = 0.01; I² = 89%) (Fig. 4). The certainty of evidence for HRQoL was rated as low, mainly due to concerns regarding risk of bias and inconsistency (Supplementary Table 7).

Fig. 4.

Forest plots: effect of digital health interventions on HRQoL

Exploratory subgroup analyses showed no statistically significant subgroup differences according to delivery modality (Chi² = 0.28, df = 1; P = 0.59; I² = 0%) (Supplementary Fig. 4) or dominant LARS management function (Chi² = 1.43, df = 1; P = 0.23; I² = 30%) (Supplementary Fig. 5). Follow-up duration was not examined as a subgroup factor because all studies included for this outcome were classified as long-term follow-up studies. Given the limited number of studies, these findings should be interpreted cautiously. Leave-one-out sensitivity analysis showed that the direction of the HRQoL effect remained consistently favorable to digital health interventions, although statistical significance was not fully stable. The pooled effect was no longer statistically significant (P = 0.09) after excluding Garfinkle et al., whereas heterogeneity was substantially reduced after excluding Zhou et al. (I² = 47%), suggesting that Zhou et al. may have contributed to the observed between-study heterogeneity (Supplementary Table 9).

The narratively synthesized HRQoL findings were mixed (Table 2). Ruan et al. [24] reported increases in QoL scores from baseline in both groups, with significantly greater improvement in the intervention group. Nehemiah et al. [26] reported significant improvements from baseline at both 12 and 24 weeks. In contrast, Kim et al. [28] found no statistically significant between-group differences in QoL outcomes.

Table 2.

Narrative synthesis of non-pooled secondary outcomes

Outcome	Instrument/measure	Study	Direction of effect	Key findings summary
Health-related quality of life (HRQoL)
HRQoL	EORTC QLQ-C30	Ruan et al. (2025)	↑	After intervention, quality of life improved in both groups, with greater improvement in the multidimensional nursing group.
	COH-QOL-CRC	Nehemiah et al. (2025)	↑	Improved from baseline at 12 weeks and at 24 weeks.
	EORTC QLQ-CR29	Kim et al. (2023)	↔	No statistically or clinically significant between-group differences.
Psychological well-being
Affect (positive/negative)	I-PANAS-SF	Sun et al. (2024)	↔	No significant differences.
Self-management & behavioral outcomes
Self-management behaviors	BSSBQ	Zhou et al. (2024)	↑	Improved more in intervention vs. control after intervention and at 3-month follow-up.
Patient engagement & activation	Motivation scale (adapted intrinsic/extrinsic)	Sun et al. (2024)	↔	No significant difference.
	PAM	Garfinkle et al. (2025)	↑	At 6 months, a higher proportion of participants increased their PAM level from baseline in the intervention vs. standard care.
Physical activity	IPAQ	Nehemiah et al. (2025)	Mixed	Physical activity increased at 12 weeks but not at 24 weeks.
Dietary quality	HEI-2015	Sun et al. (2024)	↔	No significant difference.
Social support	PSSS	Zhou et al. (2024)	↑	Improved immediately post-intervention but the effect was not sustained at 3-month follow-up.
Bowel function (non-LARS measures)
Bowel function	MSKCC-BFI	Nehemiah et al. (2025)	↑	Improved from baseline at 12 and 24 weeks.
	MSKCC-BFI	Sun et al. (2024)	↔	No significant difference.

↑, significant improvement vs. control (or vs. baseline in single-arm studies); ↔, no significant difference; Mixed, mixed findings across time points/subscales. Coding rule: For outcomes where lower scores indicate better status (e.g., LARS and HADS), reductions were coded as ↑; for outcomes where higher scores indicate better status (e.g., QoL), increases were coded as ↑. LARS Low Anterior Resection Syndrome, LARS score Low Anterior Resection Syndrome score, MSKCC-BFI Memorial Sloan Kettering Cancer Center Bowel Function Instrument, EORTC QLQ-C30 EORTC Quality of Life Questionnaire–Core 30, EORTC QLQ-CR29 EORTC Quality of Life Questionnaire–Colorectal 29, HADS Hospital Anxiety and Depression Scale, BSSBQ Bowel Symptoms Self-Management Behaviors Questionnaire, PSSS Perceived Social Support Scale, SUPPH Strategies Used by People to Promote Health, IPAQ International Physical Activity Questionnaire, HEI-2015 Healthy Eating Index-2015, I-PANAS-SF International Positive and Negative Affect Schedule–Short Form, PROMIS-SE PROMIS Self-Efficacy (Short Form 4a), GSES General Self-Efficacy Scale

Self-efficacy

Three studies evaluated self-efficacy with different instruments [24, 25, 28] and were pooled using a random-effects model, showing no statistically significant between-group difference (SMD = 0.43, 95% CI − 0.15 to 1.01; P = 0.15; I² = 79%) (Fig. 5). Subgroup analyses were not feasible because of the limited number of studies. Regarding sensitivity analyses, the leave-one-out approach showed that sequential omission of individual studies did not alter the non-significant pooled result. The pooled estimate appeared to be mainly influenced by Ruan et al., as excluding this study attenuated the effect size and eliminated statistical heterogeneity (Supplementary Table 9).

Fig. 5.

Forest plots: effect of digital health interventions on self-efficacy

Psychological outcomes

Two studies assessed anxiety and depression using the Hospital Anxiety and Depression Scale [13, 27]. No statistically significant between-group differences were observed at the reported follow-up time points (Fig. 6). One study additionally evaluated affect, with no clear between-group differences reported [25] (Table 2).

Fig. 6.

Forest plots of the meta-analyses for anxiety (A) and depression (B)

Self-management and behavioral outcomes

One randomized trial assessed self-management behaviors and found greater improvements in the intervention group than in usual care immediately after the intervention and at 3-month follow-up, suggesting a sustained behavioral benefit over the short term [12].

Two trials reported engagement-related outcomes [25, 27]. A narrative analysis was conducted due to considerable heterogeneity (I²= 96%). The trial that measured patient activation observed that a larger proportion of participants increased their activation level by 6 months in the intervention group compared with standard care, while the study found no significant between-group difference at follow-up.

One single-arm feasibility study measured physical activity and showed a short-term increase at 12 weeks, but the improvement was not maintained at 24 weeks, indicating attenuation over time [26]. Dietary quality was assessed in one randomized trial and no significant between-group difference was observed at follow-up [25]. One randomized trial evaluated social support and reported improvement immediately after the intervention, but the effect was not sustained at the 3-month follow-up [12]. Table 2 presents the detailed results of the narrative synthesis.

Bowel function

Two studies assessed bowel function using non-LARS instruments [25, 26]. Meta-analysis was not feasible because one study lacked a concurrent control group and could not provide a between-group effect estimate. Therefore, findings were synthesized narratively and were inconsistent across studies (Table 2). Nehemiah et al. [26] reported that MSKCC-BFI outcomes improved from baseline at both 12 and 24 weeks. By comparison, the randomized trial by Sun et al. [25] did not demonstrate a statistically significant between-group difference in bowel function at follow-up, including at week 26.

Other outcomes

Implementation outcomes

Five studies reported satisfaction-related outcomes. Quantitative synthesis was not considered appropriate because of methodological and outcome-reporting heterogeneity. Ruan et al. [24] and Liu et al. [11] reported satisfaction as a dichotomous outcome, whereas Nehemiah et al. [26] was a single-arm study. A pooled analysis of the remaining two randomized trials showed substantial heterogeneity (I² = 97%). Therefore, satisfaction outcomes were summarized narratively (Table 3). Higher satisfaction was observed in Garfinkle et al. [27] at 6 and 12 months and in Ruan et al. [24], where nursing satisfaction favored the intervention group. Nehemiah et al. [26] reported generally high satisfaction descriptively, whereas Moon et al. [13] and Liu et al. [11] found no clear between-group differences.

Table 3.

Narrative synthesis of non-pooled other outcomes

Outcome	Instrument/measure	Study	Direction of effect	Key findings summary
Implementation outcomes
Satisfaction	Study-specific satisfaction scale	Moon et al. (2025)	↔	No statistically significant between-group differences in satisfaction.
	Study-specific satisfaction scale	Nehemiah et al. (2025)	↑	All participants reported satisfaction with the physical-activity coaching intervention.
	Investigator-generated 2-item questionnaire	Garfinkle et al. (2025)	↑	Higher satisfaction at 6 and 12 months.
	Study-specific satisfaction scale	Ruan et al. (2025)	↑	Higher nursing satisfaction in the multidimensional nursing group.
	Satisfaction survey	Liu et al. (2021)	↔	No significant differences were reported.
Feasibility/acceptability	AIM/Adherence/retention metrics	Sun et al. (2024)	↑	AIM: Higher in intervention vs. control at Week 18 and Week 26. Adherence/Retention: Very high in both groups, with no between-group differences.
	Completion/adherence metrics	Zhou et al. (2024)	↑	High acceptability/feasibility reported (e.g., willingness to continue using the platform, completion/adherence metrics).
	usage/engagement metrics	Moon et al. (2025)	Mixed	App engagement was generally good (about two-thirds met the predefined adequate-use threshold), though some users reported access/usability issues, competing demands, or reluctance to post.
Healthcare resource utilization	study-specific questionnaire	Kim et al. (2023)	Mixed	Unplanned pharmacy visits were lower in the intervention group; other healthcare resource utilization outcomes were not significantly different.
Process/measurement outcomes
Diagnostic agreement	κ for LARS category and items	Liu et al. (2021)	↔	Agreement between WeChat-based and telephone assessments was evaluated using Cohen’s κ for LARS category and item-level responses (reported descriptively).
Follow-up efficiency	Completion rate	Liu et al. (2021)	↔	No significant differences were reported.
Knowledge (LARS-related)	Investigator-generated LARS knowledge questionnaire (7-item)	Garfinkle et al. (2025)	↑	Higher in the intervention group at 6 months and 12 months.

↑, favorable finding versus control, or versus baseline in single-arm studies; ↔, no significant difference; Mixed, mixed findings across indicators or time points. AIM Acceptability of Intervention Measure, LARS Low Anterior Resection Syndrome, κ Cohen’s kappa coefficient

Three studies contributed feasibility, acceptability, or usage data. Because Zhou et al. [12] and Moon et al. [13] reported these outcomes only among intervention participants, meta-analysis was not performed and the findings were synthesized narratively (Table 3). Sun et al. [25] showed higher acceptability scores but similarly high adherence and retention in both arms, Zhou et al. [12] reported high feasibility and acceptability based on completion and willingness-to-continue metrics, and Moon et al. [13] reported generally good engagement alongside practical barriers such as access or usability constraints and reluctance to post.

One trial assessed healthcare resource utilization, reporting fewer unplanned pharmacy visits with the intervention, while other utilization indicators did not differ clearly between groups [28] (Table 3).

Process and measurement outcomes

One study evaluated diagnostic agreement and follow-up efficiency by comparing WeChat-based versus telephone assessments: agreement for LARS category and item-level responses was described using Cohen’s kappa, and completion rates did not show meaningful differences [11]. LARS-related knowledge was assessed in one trial, with higher knowledge questionnaire scores in the intervention group at both 6 and 12 months [27]. Table 3 provides a detailed summary of the narrative findings.

Publication bias

For the primary outcome, LARS score, the funnel plot showed some dispersion among smaller studies (Fig. 7). Exploratory Egger’s test did not indicate statistically significant asymmetry (P = 0.98). A similar assessment was conducted for HRQoL. Although the funnel plot also showed some visual variability across the included studies (Fig. 8), Egger’s regression test was not statistically significant (P = 0.57). However, because these analyses were based on only a small number of studies, publication bias or small-study effects could not be excluded.

Fig. 7.

Funnel plots of the LARS score

Fig. 8.

Funnel plots of the HRQOL

Discussion

Symptom management and long-term functional recovery remain central challenges for patients experiencing LARS after sphincter-preserving rectal surgery. In this systematic review, we synthesized quantitative and narratively summarized findings on digital health interventions designed to support LARS management. Overall, the body of evidence suggests that technology-enabled care may alleviate LARS symptom burden and confer benefits for health-related quality of life, although the certainty of evidence varies across outcomes and several secondary domains remain insufficiently characterized. Taken together, these findings suggest that digital health approaches have promising potential to extend structured LARS self-management beyond routine postoperative follow-up.

LARS score

This review suggests that digital health interventions may reduce overall LARS symptom burden compared with usual care. This finding should be interpreted as suggesting that structured and continuous LARS-related support delivered through digital or remote approaches may be beneficial, rather than that digital technology itself is independently effective. Most included interventions shared core elements, such as patient education, symptom monitoring, remote follow-up, dietary or lifestyle guidance, and professional support, which are consistent with multimodal LARS management [4].

From a mechanistic perspective, digital interventions may be beneficial because they can deliver a structured package of education, repeated symptom assessment, and responsive feedback throughout the recovery trajectory [29]. This approach may enable earlier recognition of symptom deterioration and more timely adjustment of self-management strategies, such as dietary modification, medication optimization, pelvic floor training, coping and problem-solving, and appropriate escalation of care [30]. Given that LARS symptoms often fluctuate and that embarrassment or stigma can delay help-seeking, remote and structured touchpoints may lower access barriers and facilitate proactive management [31, 32].

The exploratory subgroup analyses showed no significant differences according to delivery modality, dominant management function, or follow-up duration. However, this should not be interpreted as evidence that different digital approaches are equivalent, because the number of studies within each subgroup was limited and several interventions included overlapping components. The sensitivity analysis further supported the stability of the finding, although the reduction in heterogeneity after excluding one study suggests that differences in intervention design, study population, or follow-up assessment may have contributed to between-study variability [13].

Overall, the findings are promising but remain preliminary. The certainty of evidence for LARS score was moderate because of concerns regarding risk of bias and imprecision, and inconsistent outcome reporting limited comparability across studies. Future well-designed randomized controlled trials with standardized LARS outcome reporting and clearer descriptions of intervention components are needed.

Health-related quality of life

This review suggests that digital health interventions may enhance health-related quality of life in patients with LARS. This finding is consistent with recent randomized evidence indicating that digital programs can improve global HRQoL [33]. A likely mechanism is that structured digital education combined with repeated symptom monitoring and responsive feedback can improve day-to-day symptom control and perceived coping capacity, which may translate into better overall well-being [10].

However, this finding should be interpreted cautiously. Sensitivity analyses showed that the direction of effect remained generally favorable, although statistical significance and heterogeneity were influenced by the exclusion of individual studies. The included studies used different HRQoL instruments and often reported selected domains rather than overall scores, which limited cross-study comparability and quantitative pooling. Exploratory subgroup analyses did not suggest clear differences according to delivery modality or dominant management function, but these results were limited by the small number of studies within each subgroup. Future trials should prespecify the quality-of-life construct of interest, report standardized global scores alongside key domains at aligned follow-up time points, and use consistent reporting formats to enable synthesis.

Self-efficacy & psychological outcomes

Current evidence does not show a clear benefit of digital health interventions for self-efficacy or broader psychological outcomes. The pooled effect on self-efficacy was not statistically significant and heterogeneity was substantial. Trials reporting anxiety and depression using the Hospital Anxiety and Depression Scale and those assessing affect also found no meaningful between-group differences.

These null findings may reflect outcome–intervention misalignment rather than true absence of benefit. Most interventions primarily targeted symptom education, repeated monitoring, and behavioral self-management. By comparison, generalized self-efficacy and emotional distress are influenced by multiple concurrent factors across rectal cancer survivorship, including treatment sequelae, functional limitations, comorbidity, and psychosocial stressors, and therefore may change more slowly or less directly than symptom burden [34]. This finding is consistent with a prior meta-analysis in colorectal cancer survivors reporting no clear improvement in psychological outcomes despite behavioral gains [35]. In addition, few trials, modest sample sizes, variable follow-up windows, and the use of relatively generic mental health measures may have reduced sensitivity to change and diluted detectable effects.

The substantial heterogeneity in self-efficacy estimates is plausibly driven by differences in what was measured and how it was targeted. Across studies, self-efficacy may have been operationalized as general confidence in coping rather than confidence in managing bowel symptoms, and intervention intensity and responsiveness likely varied, leading to divergent effect sizes. Differences in postoperative timing and baseline symptom severity may further modify responsiveness. Sensitivity analyses suggested that heterogeneity was largely driven by a single study, indicating that study-level design and intervention characteristics may explain much of the variability [24].

Future trials should prespecify whether the intended target is general self-efficacy, bowel-specific self-efficacy, or psychological distress, select measures that match the construct and capture bowel-related psychosocial burden, and align assessment time points across studies with follow-up durations adequate to detect mental health change. Interventions may also need to incorporate explicit psychosocial elements in addition to symptom management, such as structured coping-skills training and escalation pathways for clinically significant distress, to meaningfully shift psychological outcomes.

Self-management and behavioral outcomes

Evidence for self-management and behavioral outcomes was encouraging but too sparse and inconsistently reported to support firm conclusions. Only one randomized trial assessed self-management behaviors and suggested a short-term benefit [12], whereas LARS-related knowledge was evaluated in a single study and appeared higher at later follow-up, implying that structured digital education may strengthen condition-specific understanding over time [27]. A meta-analysis supported the plausibility that structured digital education can strengthen self-management capacity over time [33]. Activation and engagement findings were particularly variable, with divergent results and substantial heterogeneity that likely reflects construct and measurement misalignment as well as differences in intervention intensity and use. Because activation is shaped by baseline readiness and real-world constraints on acting upon guidance, and because behavioral change is strongly engagement-dependent, incomplete uptake of educational and monitoring components may further attenuate effects [36]. Future trials should prespecify the intended pathway construct, align intervention components with matched, validated measures, standardize follow-up windows, and report engagement and fidelity in sufficient detail to clarify dose–response patterns and key active ingredients.

Other outcomes and real-world applicability

Implementation evidence remains relatively limited, which constrains inferences about scalability and routine adoption. Although digital interventions are often positioned as convenient extensions of follow-up care, real-world effectiveness is likely to depend on whether the program is usable, acceptable, and sustainable for patients over time, and whether clinical teams can integrate the workflow into existing postoperative pathways [37]. Key practical considerations include the clarity of triage and escalation procedures for symptom deterioration, the extent of professional support required to deliver feedback, and the resources needed for onboarding, training, and ongoing technical assistance [38].

Engagement is a central determinant of impact in real-world settings [36]. In routine care, patients may experience fluctuating motivation, competing treatment demands, and varying digital literacy, all of which can influence adherence to symptom reporting, educational modules, and recommended self-management behaviors [39]. Digital approaches also raise equity considerations: differential access to smartphones, data plans, and support may widen disparities if interventions are not designed with inclusive usability and low-burden participation in mind [40]. Future studies should therefore report implementation outcomes more systematically, including feasibility, acceptability, usability, reach, uptake, adherence trajectories, fidelity of delivery, and reasons for non-use, alongside pragmatic indicators such as staffing time, workflow integration, and resource utilization [41]. This will be essential to determine which digital models are effective, feasible, and scalable in routine care.

Limitations of the study

First, the findings of this review were limited by the quality, size, and design of the included studies. The current evidence was based on a small number of studies, many of which had modest sample sizes and some methodological limitations. Importantly, the evidence base included both randomized and non-randomized studies, including one observational cohort study, which introduced methodological heterogeneity and may have affected the reliability of the pooled estimates. Non-randomized studies are more susceptible to selection bias, baseline imbalance, and unmeasured confounding, which may have influenced the estimated intervention effects. Several studies had concerns related to outcome assessment, missing data, or incomplete reporting, which may have introduced bias and reduced confidence in the findings. Furthermore, differences in intervention content, delivery modality, intensity, follow-up duration, and outcome measures contributed to clinical and methodological heterogeneity, thereby limiting the comparability and interpretability of findings across studies.

Second, limitations of the review process should be acknowledged. Although we searched multiple databases, selected gray literature sources, trial registries, and reference lists, the review was restricted to studies published in English or Chinese. Therefore, relevant studies published in other languages, indexed in other databases, or available only as unpublished or difficult-to-access gray literature may have been missed. Despite efforts to develop a comprehensive search strategy, the possibility of missing eligible studies cannot be completely excluded.

Third, the meta-analyses were limited by the small number of studies contributing to each pooled estimate. For several outcomes, quantitative synthesis was not feasible because of differences in study design, inconsistent outcome reporting, or the absence of a concurrent control group. Among the outcomes that were pooled, substantial statistical heterogeneity was observed for some secondary outcomes, particularly health-related quality of life and self-efficacy. The use of different measurement instruments and effect measures may also have reduced the interpretability of pooled estimates. In addition, because only a limited number of studies were available, publication bias and small-study effects could not be ruled out.

Overall, these limitations should be considered when interpreting the findings and their generalizability. The current evidence suggests that digital health interventions may reduce LARS symptom burden and may improve health-related quality of life, but the certainty of these conclusions remains limited. The findings are most applicable to patients with LARS after sphincter-preserving rectal surgery who receive digitally supported education, symptom monitoring, self-management guidance, or remote follow-up. However, their applicability to broader postoperative populations, different healthcare systems, and routine clinical settings remains uncertain. Further adequately powered and rigorously designed studies with standardized outcome measures, prespecified follow-up time points, and clearer reporting of intervention components and implementation processes are needed before firm clinical recommendations can be made.

Implications for practice and research

Digital health interventions may be considered as a complementary approach to routine postoperative follow-up for patients experiencing, or at risk of, LARS. By providing structured education, ongoing symptom check-ins, and timely support between clinic visits, these programs may enable earlier identification of symptom deterioration, reinforce self-management skills, and support more responsive care over the recovery trajectory. Their practical value is likely to be greatest when embedded within a defined clinical pathway that specifies how symptom reports trigger professional review, triage, escalation, and adjustments in management, with explicit attention to workload, documentation, and coordination with existing follow-up schedules [42]. Implementation should prioritize feasibility and equity by minimizing patient burden, using accessible interfaces, and providing onboarding, troubleshooting, and alternative contact options so that patients with limited digital access, variable digital literacy, or additional social needs are not excluded [43].

Future research should move beyond demonstrating potential benefit to clarifying what works, for whom, and under what implementation conditions. Trials should adopt harmonized outcome definitions and standardized reporting for LARS and related patient-centered outcomes with prespecified time points, ideally supported by a core outcome set to improve comparability and reduce selective reporting. Intervention reporting should describe components, delivery, intensity, and hypothesized mechanisms in sufficient detail to enable replication, alongside engagement and adherence trajectories, reasons for non-use, and fidelity indicators on both patient and clinician sides. Adequately powered multicenter and pragmatic designs are needed to examine effect modifiers such as baseline LARS severity, postoperative stage, comorbidity burden, and digital literacy, and to test whether responsive feedback loops and escalation pathways add value beyond education alone by linking intermediate outcomes to downstream clinical endpoints. Implementation and health system outcomes should be integrated as core endpoints, including feasibility, acceptability, reach, adoption, sustainability, staffing requirements, resource utilization, economic impact, and equity.

Conclusions

In conclusion, this review suggests that digital health interventions have promising potential to reduce overall LARS symptom burden and may improve health-related quality of life in patients undergoing sphincter-preserving rectal surgery during postoperative recovery. Although the overall findings indicate clinically meaningful potential, the small number of available trials, clinical and methodological heterogeneity, and incomplete harmonization of outcome definitions and reporting require that these results be interpreted as preliminary and with caution. Future digital LARS care models should be tailored to patients’ symptom profiles and recovery phase, and embedded within clear pathways for symptom monitoring, triage, escalation, and timely feedback. Priority research directions include adequately powered trials with standardized outcome sets and aligned reporting time points, longitudinal studies evaluating durability of effects beyond early recovery, and mechanism- and implementation-informed evaluations that clarify active components, engagement requirements, workflow feasibility, and equity implications.

Abbreviations

CI

Confidence Interval

I²

Inconsistency Index

ITT

Intention-to-Treat

LARS

Low Anterior Resection Syndrome

mHealth

Mobile Health

MD

Mean Difference

RCT

Randomized Controlled Trial

RevMan

Review Manager

RoB 2

Risk of Bias Tool Version 2

ROBINS-I

Risk Of Bias In Non-randomized Studies of Interventions

SD

Standard Deviation

SMD

Standardized Mean Difference

Authors’ contributions

ZYY, CXY, and NL contributed to the study design, data collection, data analysis and interpretation, and drafting of the manuscript. YZJ contributed to data interpretation and critically reviewed the manuscript for important intellectual content. WJZ conceptualized and supervised the study. All authors have read and approved this manuscript. All authors read and approved the final manuscript and agreed to be accountable for all aspects of the work.

Funding

This work was supported by the Shanxi Nursing Association (Key Project; grant no. SXHLKY-202502) and the Shanxi Medical University (2024 Education and Teaching Research Project; grant no. XJ2024103). The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.

Data availability

All data generated and analyzed are included in the article, Supplementary Materials and are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable. This manuscript is a systematic review and meta-analysis, which analyzes data extracted from publicly available literature. It does not involve any new human participants, human data, human tissue, or animals. The review protocol was registered in PROSPERO under registration number CRD420251250663, in accordance with best practices for systematic reviews and to ensure transparency and reproducibility.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.