A total of 7906 records were initially identified through database searching. After deleting duplicates, a total of 4781 studies were selected. Finally, a total of 10 studies were included in the review, with a total of 1835 participants analyzed. The PRISMA flowchart is depicted in Figure 1.
The characteristics of the studies included are presented in Table 1.
All included studies were published between 2011 and 2020. Mean ages of participants ranged from 44.6  to 59.18 years . The percentage of women were higher in all studies (79.2-100%), even including only women some of them [34,36,37,40]. In regard to the cancer etiology, most studies included only patients with breast cancer [34-42]. The cancer treatment included surgery, radiotherapy, chemotherapy, hormone therapy or a combination thereof. Table 1 also shows the modified Downs and Black scale scores. The mean score of the studies included was 20.1. Based on the cut points suggested to categorize studies according to their quality, three articles were evaluated as ‘‘fair’’ (15–19 points) and seven were classified as ‘‘good’’.
The risk of bias assessment using the Cochrane Risk of Bias Tool for randomized trials is presented in Table 2.
The main characteristics of the studies included are shown in Table 3.
Tabla 3 includes the intervention type, the approach of the comparator group, the eHealth system used, the duration, the physical variables included and the main results.
Comparator group approach
Six studies compared the eHealth intervention with a comparator group who did not received any intervention [33,37,38,40-42]. Three studies compared the eHealth physical activity intervention with a conventional treatment (information , brochure , traditional physical activity ) and one study compared the eHealth physical intervention with a diet intervention .
Most studies included used a mobile app to carry out the physical activity intervention [34,36,39]. Two studies performed the physical activity intervention using interactive emails [35,38]. Two studies used a website to instruct and provides personalized feedback to patients [33,42]. One study carried out the physical activity intervention using the Fitbit app or website, and adding regular emails and phone calls . One study used an online sheet to increase physical activity and other study carried out telephone counseling sessions to guide the physical activity intervention . Feedback during physical activity intervention was used in seven of the included studies [33,34,37,39-42].
Duration of the intervention
Details of the intervention duration for each study are also presented in Table 3. Median intervention length was 3.2 months (range 1–6 months). The majority of the studies performed a three months intervention [35,37-41], two studies carried out an intervention during 6 months [33,34] and two studies performed a 4 weeks treatment [36,42]. No specific prescriptions about physical activity frequency were provided in most studies.
Physical activity variables
The included studies evaluated PA using different tools, e.g. using the International Physical Activity Questionnaire (IPAQ) [37,39,42], the Short Questionnaire to Assess Health Enhancing Physical Activity (SQUASH) , the Godin-Shepard Leisure-Time Physical Activity questionnaire [34,40], the Physical Activity Questionnaire (PAQ) , the Behavioral Risk Factor Surveillance System (BRFSS) , the ActiGraph wGT3X-BT [36,41] or the FitBit One . The most reported outcome was the IPAQ, followed by the Godin-Shepard Questionnaire and the ActiGraph wGT3X-BT.
Results obtained in meta-analysis
Data from 9 RCTs were included [33-40,42]. Excluded studies did not provide (sufficient) physical activity data (either baseline and/or post intervention means and standard deviations) and attempts to contact trial authors were unsuccessful. The analysis was based on 1314 patients (658 for intervention and 656 for control).
For the physical activity level, the pooled mean difference (MD) showed significant overall effect when compared the eHealth intervention to no intervention (MD=1.14; 95% CI=0.18, 2.10; p=0.02; five studies [33,37,38,40,42]), to a conventional treatment (MD=0.20; 95% CI=0.02; 0.38; p=0.03; three studies [34,36,39]) or to a diet intervention (MD=0.19; 95% CI=0.68; 1.70; p<0.001, one study ). Heterogeneity was high in the comparison with no intervention and conventional treatment (I2=96%; I2=93%), respectively.
These interventions resulted in an overall effect of 0.83 (95% CI 0.35–1.31), which was significant (Z = 3.38, p<0.001). There was significant (Chi2 = 15.78; p<0.001) and high heterogeneity (I2 = 87.3%) across the included comparisons.