Search results and selection
Following an initial search of electronic databases, 2,290 articles were retrieved. After removing 1,029 duplicates, 996 articles were excluded based on their titles and abstracts as they were irrelevant to the topic, leaving 33 documents for a full-text review. Nine documents [28, 29, 37,38,39,40,41,42,43] met the inclusion criteria. The remaining 24 studies were excluded, and the specific reasons for their exclusion are provided in Supplementary Material 2. The screening process is illustrated in Fig. 1.
Flow chart of study selection
Description of included studies
Study characteristics
The characteristics of the included studies are summarized in Table 2. These studies were conducted in five countries and were published between 2015 and 2024: Sweden (two) [29, 39], China (four) [37, 40, 41, 43], the Netherlands (one) [38], South Korea (one) [42] and Iran (one) [28]. Three of these studies referred to theoretical intervention models, namely the Sound Heart Model [28], the Cognitive Behaviour Therapy Model (CBT) [29, 38], and Acceptance and Commitment Therapy (ACT) [38]. Four studies performed follow-up assessments [38, 39, 42, 43].
Characteristics of parents
A total of 727 parents of children with cancer participated in the studies; the number of participants per study ranged from 21 [39] to 146 [37]. The parents varied between 31.0 and 49.9 years. The types of paediatric cancer included acute leukaemia, osteosarcoma, eye tumours, neurological cancers, lymphoma, Wilms’ tumour, hematologic cancer and solid tumour.
Characteristics of digital health interventions
The intervention duration ranged from four weeks [42] to six months [37, 40]. Researchers in five studies delivered interventions via websites. Asadzandi et al. (2020) [28] implemented an intervention based on the Sound Heart Model, which used web-based educational software to foster optimism, hope, and courage in parents of children with cancer to enhance their mental health. Cernvall et al. (2015) [29] conducted an Internet-based guided self-help intervention program that taught parents of children with cancer coping strategies for psychological stress through an online course. Joosten et al. (2024) [38] used Op Koers Online to teach positive coping skills through chat sessions to help parents of children with cancer reduce or prevent mental health problems. Sveen et al. (2021) [39] provided Internet-based cognitive behavioural therapy for parents who have lost a child to cancer to manage insomnia. In addition, Park et al. (2023) [42] implemented an internet-based family resilience-promoting program, focusing on emotional coping, family roles, communication, and problem-solving. Researchers in four other studies have used an APP to deliver interventions. Wu et al. (2022) [40] provided disease information, online education, and relevant counselling to the parents of children with osteosarcoma based on a We-Chat platform. After two years, Duan et al. (2024) [37] similarly used the We-Chat platform to offer a comparable intervention to parents of children with acute lymphoblastic leukaemia. Wang et al. (2024) [41] delivered a 12-week WeChat-based education, relaxing, and care program (WERC) to the parents of children with lymphoma. Luo et al. (2021) [43] delivered an 8-week mobile device-based resilience training program via WeChat, focusing on enhancing resilience through weekly sessions on relaxation, problem-solving, cognitive strategies, and communication skills for parents of children with cancer.
Characteristics of controls
All participants in the control group received usual care. Additionally, two studies [29, 38] permitted parents in the control group to use appropriate digital health devices after the study ended.
Outcome measures
Three measurements were used to evaluate the PTSD of participants: the PTSD Checklist Civilian Version (PCL-C) [29], the PTSD Checklist for DSM-5 (PCL-5) [39] and the Impact of Event Scale-Revised (IES-R) [40]. Six tools were applied to assess the participants’ anxiety: Depression, Anxiety, and Stress Scale (DASS 21) [28], Beck Anxiety Inventory (BAI) [29], Self-rating Anxiety Scale (SAS) [37, 41], Patient-reported Outcomes Measurement Information System [38], Generalized Anxiety Disorder-7 (GAD-7) [39] and Hospital Anxiety and Depression Scale (HADS) [40]. Six tools were utilized to assess depression: Depression, Anxiety, and Stress Scale (DASS 21) [28], Beck Depression Inventory-II(BDI-II) [29], Self-rating Depression Scale (SDS) [37, 41], Patient-reported Outcomes Measurement Information System [38], Montgomery-Asberg Depression Rating Scale (MADRS) [39] and Hospital Anxiety and Depression Scale (HADS) [40], Beck Depression Scale [42], Self-Rating Depression Scale [43]. The relevant outcome measures (PTSD, anxiety, and depression) were assessed using standardised and extensively validated scales. These instruments have been repeatedly validated, demonstrating good reliability and validity. Therefore, they can be regarded as reliable tools for use in clinical practice.
Risk of bias
The methodological quality of the included studies was presented in Fig. 2. In the randomization process (D1), the study by Asadzandi et al. (2020) [28] was assessed as “some concerns” due to insufficient details regarding the randomization method. For dimensions such as deviations from the intended interventions (D2), missing outcome data (D3), and selection of the reported result (D5), all nine studies showed a low risk of bias. Notably, regarding measurement of the outcome (D4), all studies were rated as “some concerns.” This was primarily due to the subjective nature of the assessment scales used in the studies, which are easily influenced by the participants’ subjective judgment. Additionally, the specific characteristics of digital health interventions, such as the recognizability of the intervention, made it impossible to implement blinding of participants. The interplay of these two factors may introduce a risk of measurement bias. Therefore, the overall rating of risk of bias for all studies was “some concerns.”
Risk of bias assessment for the included studies
Meta-analysis results
PTSD
Four studies assessing PTSD were included for meta-analysis [29, 39,40,41]. Since there was low heterogeneity among the studies (I2 = 0%, P = 0.53), they were combined using a fixed-effects model. The results indicated a significant difference between the experimental and control groups [SMD= −0.37, 95%CI −0.61, −0.13, P = 0.003], see Fig. 3(A).
Subgroup analysis results were as follows: (1) Intervention type: two studies conducted interventions using websites [29, 39], while the other two implemented interventions using apps [40, 41]. The results from both subgroups indicated that the intervention effects in the experimental group were more effective than those in the control group [SMD = −0.51, 95% CI −0.97, −0.04, P = 0.03], [SMD = −0.32, 95% CI −0.60, −0.04, P = 0.03], as shown in Fig. 4(A). Additionally, no significant heterogeneity was observed between the subgroups (I2 = 0%). (2) Intervention duration: the intervention duration for two studies was ≤ 10 weeks [29, 39], while the remaining studies had an intervention duration of >10 weeks [40, 41]. This subgroup analysis also showed that the intervention effects in the experimental group were more effective than those in the control group [SMD = −0.51, 95% CI −0.97, −0.04, P = 0.03], [SMD = −0.32, 95% CI −0.60, −0.04, P = 0.03] (Fig. 5(A)). Furthermore, no substantial heterogeneity was observed between the subgroups (I2 = 0%).
Forest plot of subgroup analysis by intervention type
Forest plot of subgroup analysis by intervention duration
The sensitivity analysis showed that the results were stable. (Supplementary Material 3, Table S1. Sensitivity Analysis Results for Post-Traumatic Stress Disorder Using the Leave-One-Out Method (Fixed-effects Model)).
Anxiety
Seven studies were included in the meta-analysis [28, 29, 37,38,39,40,41] and owing to moderate heterogeneity between studies (I2 = 45%, P = 0.09), we used a random-effects model. A statistically significant difference was observed between digital health interventions and usual care in reducing anxiety symptoms [SMD= −0.42, 95%CI −0.65, −0.18, P< 0.001], see Fig. 3(B).
Subgroup analysis results were as follows: (1) Intervention type: four studies conducted website-based interventions [28, 29, 38, 39], while three studies used an app format for the intervention [37, 40, 41]. The findings indicated that there were no significant differences in intervention effects between the experimental and control groups in the website group [SMD = −0.39, 95% CI −0.92, 0.14, P = 0.15], while in the app group, the experimental group showed significantly better intervention effects than the control group [SMD = −0.41, 95% CI −0.62, −0.19, P < 0.001], as shown in Fig. 4(B). However, no statistically significant differences were found between subgroups (P = 0.95), with no evidence of heterogeneity (I²=0%). (2) Intervention duration: three studies had an intervention duration of ≤ 10 weeks [29, 38, 39], while four studies had an intervention duration of >10 weeks [28, 37, 40, 41]. In subgroups with intervention durations of < 10 weeks, there was no significant difference in intervention effects between the experimental and control groups [SMD = −0.17, 95% CI −0.49, 0.14, P = 0.27]. However, in subgroups with intervention durations of >10 weeks, the experimental group showed significantly better intervention effects than the control group [SMD = −0.55, 95% CI −0.86, −0.25, P < 0.001], as shown in Fig. 5(B). Nevertheless, the between-subgroup differences were not statistically significant (P = 0.09). Additionally, substantial heterogeneity was observed between the subgroups (I2 = 66.1%).
Sensitivity analysis demonstrated that individual trials did not change the results (Supplementary Material 3, Table S2. Sensitivity Analysis Results for Anxiety Using the Leave-One-Out Method (Random-effects model)).
Depression
Nine studies were included in the meta-analysis [28, 29, 37,38,39,40,41,42,43]. Due to low heterogeneity between studies (I2 = 30%, p = 0.18), we used a fixed-effects model. The results showed that digital health interventions can improve depression symptoms, and the difference was statistically significant [SMD= −0.47, 95% CI −0.62, −0.32, P<0.001], see Fig. 3(C). Subgroup analysis results were as follows: (1) Intervention type: five studies implemented website-based interventions [28, 29, 38, 39, 42], while four other studies used an app format for the intervention [37, 40, 41, 43]. The results showed that in both the website group and the app group, the experimental group demonstrated superior intervention effects compared to the control group [SMD = −0.59, 95% CI −0.98, −0.20, P = 0.003], [SMD = −0.40, 95% CI −0.59, −0.21, P < 0.001], as shown in Fig. 4(C). Additionally, no significant heterogeneity was observed between the subgroups (I2 = 0%). (2) Intervention duration: the intervention duration for five studies was ≤ 10 weeks [29, 38, 39, 42, 43], while the remaining studies had an intervention duration of >10 weeks [28, 37, 40, 41]. Both subgroups showed that the experimental group demonstrated superior intervention effects compared to the control group [SMD = −0.44, 95% CI −0.67, −0.21, P < 0.001], [SMD = −0.54, 95% CI −0.89, −0.19, P = 0.002], as shown in Fig. 5(C). Furthermore, no substantial heterogeneity was observed between the subgroups (I2 = 0%).
Sensitivity analysis revealed that individual studies did not alter the overall results. (Supplementary Material 3, Table S3. Sensitivity Analysis Results for Depression Using the Leave-One-Out Method (Fixed-effects Model)).
link