To identify potentially relevant studies, CINAHL, Embase, PsycINFO, Cochrane Network, and PubMed databases were searched for eligible RCTs from January 2013 to January 2020. The review began in 2018. Due to the rapid development of medical technology in health care, only studies from the last 5 years were included. During the time of writing this paper, the search was continually updated until January 2020, while maintaining the years 2013 and 2014, given the relevance of the included studies. The search strategy consisted of Medical Subject Headings combined with text words for cancer (Textbox 1) in a Boolean search. A medical information specialist checked the final search syntaxes. DJMA and MJHMVB screened the titles, abstracts, and full texts. Interresearcher reliability was checked using a 20% random sample of abstracts and full texts. Consensus was reached through discussion.

The following information was extracted from each publication: study characteristics (country of origin, year of publication, aim, and sample size), patient characteristics (age, gender, and type of disease), intervention characteristics (content, duration, and frequency), and outcome measures (instruments used and effects on physical and psychosocial outcomes). The first author independently extracted the data, and another author checked the data extraction for 20% of the studies to determine interrater reliability. Consensus was reached through discussion.

The methodological quality of the studies was evaluated, but it did not serve as an eligibility criterion. We used the CONSORT (Consolidated Standards of Reporting Trials) list developed by the CONSORT group to identify the problems arising from inadequate reporting of RCTs [17]. Items were scored using a tick mark. The tick marks indicate “yes” as an answer to the question, resulting in a maximum quality score of 37. For the qualitative synthesis, we counted the overall scores and classified them into 3 quality categories: high quality (CONSORT score >25), moderate quality (CONSORT score 13-25), and low quality. (CONSORT score <13)

Two reviewers (DJMA and MJHMVB) independently reviewed the papers and independently assessed the methodological quality. In case of disagreement, consensus was reached through discussion.

Figure 1 outlines the search process. A total of 6047 references were identified through the search. Screening titles, abstracts, and full texts yielded 19 eligible studies.

A total of 19 publications were included (Figure 1); 3 publications were based on the same study, assessing different outcome measures [18-20]. The sample sizes ranged from 39 to 752 patients. All studies had a pre- and posttest design to measure the outcome differences in the groups. One study used 2 experimental groups [21], examining an internet-based patient-provider communication service with or without the additional use of a web-based illness management system, and another study used 2 experimental groups [22]: an unsupervised group that used a mobile app to record data or a supervised group that used the app and reviewed data with a physician. All other studies used a single experimental and control group (eg, control group assigned to a waiting list and received the intervention after the active treatment group, control group receiving written formats, and control group receiving usual care).

Table 1 presents the methodological quality of the studies. Allocation concealment was described in 9 studies [18-26]. Blinding of participants, personnel, and outcome assessors was adequately described in 8 studies [18,19,22,24,26-29]. Two studies explicitly described a nonblinded approach [21,30]. Out of 37 points, one study achieved 33 (89%) points and had the highest score [29]. A total of 13 studies were of high quality and 6 were of moderate quality [31].

The 19 studies included 5186 patients. Eleven studies included patients in the active treatment phase [21-25,27,28,33-36]. Eight studies included patients who had finished active treatment and were in the curative setting, in the follow-up phase, or in the palliative setting, in the supportive care phase [18-20,26,29,30,32,37]. Nine studies were related to DSMSTs for patients with breast cancer [21,22,24-27,29,32,34]. Six studies were related to patients with cancer in general [18-20,23,33,37]. Two studies included 129 newly diagnosed patients with cancer, of whom 92 were treated for breast cancer [35], and 625 cancer survivors, of which 138 were treated for breast cancer [30]. Two breast cancer studies focused on patients undergoing chemotherapy [22,24]. Of the remaining studies, 1 study focused on 261 patients with primary cancers that had metastasized to the liver and 1 study on 285 patients with nonsmall cell lung cancer [28,36]. The mean number of participants was 273 (range 39-752), of which 70.99% (3682/5186) were female. Some studies included only female participants. The mean age of the subjects was 54.2 years (range 42.35-61.7; Table 2).

The intervention characteristics for both the intervention and control groups are described in Table 2. The degree of detail provided about the interventions varied greatly across studies. There was a large variation in the duration, frequency, and content of the interventions. The mean duration of the intervention was 39.5 weeks (range 2 weeks to 25 months). A total of 37% (7/19) interventions focused only on the psychological well-being of patients [27,29,30,33-36], 1 focused only on physical health [32], and 11 focused on both [18-26,28,37].

The technological component was mainly a web-based approach (16/19, 84%) [18-21,23,25-30,32,33,35-37]; in 2 studies, a mobile app was used [22,34], and 1 study used a mobile game as a DSMST [24]. Of the 16 studies that used a web-based approach, 1 study sent email reminders in an attempt to maintain or improve adherence [29].

Although the content of the interventions differed, 5 key components of DSMSTs were identified (Table 2). A total of 5 of the 16 web-based approach studies used all 5 key components in their DSMSTs [21,25,28,29,32,37] to increase self-management (support): (1) An assessment component to monitor and report symptoms, problems, and priorities for support along physical, functional, and psychosocial dimensions, currently and over time (eg, improving diet, increasing exercise, and stress management via relaxation therapy); (2) Tailored symptom self-management support to self-manage symptoms and problems the patient experiences (eg, in the study by Børøsund et al [21], patients could choose symptoms and problems they were experiencing from a predefined list, rate the burden of these symptoms and problems, and indicate where they needed help); (3) The information section, which included information about various aspects of cancer such as exercise, nutrition, coping, and symptom management and also provided access to other reliable and relevant web sources; (4) Communication section, with fellow patients or with health care providers, using discussion centers, an SMS text messaging function, or email as a communication tool. Communication with fellow patients was often used for social networking, providing feedback, and encouraging each other, whereas communication with health care providers was often used for difficult questions and support; (5) Diary, where patients could keep personal notes. The 2 studies that provided a mobile app only offered 1 of the 5 key components. One offered tailored information and the other offered an assessment component (Table 2) [22,34]. One study, using a mobile game, offered patient education as a key component to increase the self-management of patients with breast cancer [24].

The measurement instruments used and the corresponding outcomes of the studies are presented in Table 3. Psychosocial outcome measures, such as QOL, anxiety and depression, and symptom distress, were the most commonly used outcome measures, mostly using validated (eg, The European Organisation for Research and Treatment of Cancer QOL Questionnaire Core 30, Functional Assessment of Cancer Therapy-Fatigue, Hospital Anxiety and Depression Scale [HADS]) questionnaires.

A total of 10 studies reported QOL [20,23-28,30,32,35], whereas 6 of them reported positive outcomes [24,26,28,30,32,35]. A few studies (4/9, 44%) used The European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 to measure QOL. Four studies observed that the overall health-related QOL improved to a significantly larger degree compared with the control group [24,28,30,35]. Two of these studies [24,30] were of high quality, including 76 and 625 participants, respectively. One study, including 81 patients who finished active treatment, with 8 weeks of access to an internet-based tailored exercise program, found that health-related QOL improved to a significantly larger degree compared with the control group on the subdomains global health status, physical functioning, role functioning, cognitive functioning, and arm symptoms [26]. One study, which reported on 59 patients with breast cancer who finished active treatment, only found a statistically significant difference in the subdomain physical functioning [32]. Four studies [20,23,25,27] found no statistically significant differences in the overall QOL or subdomains of QOL. Three of these studies included >300 patients each, of which 2 studies were of high quality.

Anxiety was reported in 5 studies, whereas depression was reported in 8 studies. Four studies [21,24,34,35] reported anxiety and depression. Four studies used the HADS [20,21,34,35] to report anxiety and depression. Willems et al [20] used only the depression subscales of the HADS. Two other studies also reported depression using different questionnaires [28,37]. One study only reported the outcomes of anxiety [27]. Two studies reported significant differences in anxiety levels [21,34], and 3 studies reported significant differences in depression levels [21,28,34]. One of these studies, including 167 patients with breast cancer in active treatment with access to a web-based self-management support system with e-messages, observed significantly lower anxiety (P=.03) and depression (P=.03) levels in the intervention group than in the usual care group [21]. This study was classified as high quality. Another study, classified as moderate quality and including 261 active treatment patients, reported a reduction in depression (Cohen d=0.71) for the intervention group when compared with the usual care group [28]. In contrast, another moderate quality study, including 39 patients with breast cancer with access to a mobile app for 2 weeks, reported significantly lower anxiety (P=.02) and depression (P=.03) levels in the control group than in the intervention group [34]. Of the studies that found no significant differences in anxiety, 2 studies were of moderate quality, whereas 1 was classified as high quality, including 76 patients who completed active cancer treatment. Of the studies that found no significant differences in depression, 2 studies were of moderate quality, whereas 3 [20,24,25] were classified as high quality. One of these studies included 76 patients who completed active cancer treatment and 2 of these studies included 325 patients, during their active cancer treatment.

A total of 7 studies reported symptom distress [21,23,25,29,33,35,36], of which 6 were classified as high-quality studies. One study [35] was classified as moderate quality. All studies used a web-based approach. One study (including 167 active treatment patients), using 2 experimental groups [26] and examining an internet-based patient-provider communication service with and without the additional use of a web-based illness management system, found significantly lower symptom distress in the web-choice intervention group than in the control group, but no statistically significant differences were observed between the 2 intervention groups [21]. In addition, 2 other studies, including 150 and 752 patients, found significantly less distress in the intervention group [29,33]. Another study, including 325 patients, found significant group differences in symptom distress [25]. One study included 285 active treatment patients and their primary caregivers [36]. These caregivers reported lower patient physical symptom distress in the intervention group than in the control group. One study found that distress was significantly lower immediately after the intervention in the intervention group than in the control group. However, distress did not change significantly from immediately after the intervention to the follow-up 2 months later [35]. One study, including 60 patients, found no statistically significant group×time interactions [23].

Five studies reported fatigue [20,26,28,32,37], of which 2 reported a decrease in fatigue after 8 and 12 weeks of a web-based intervention [26,32]. One of these studies, including 81 patients, was classified as high quality. Three others found no significant changes after an intervention of 6 months [20,28,37], of which 1 was classified as high quality, including 462 patients.

Six studies reported results on PA [18,19,22,23,32,37]. Four studies observed significant effects; 2 studies were classified as high quality [19,23], whereas 2 were classified as moderate quality [32,37]. In 1 study, 139 participants were randomly assigned to an unsupervised group (intervention), a supervised group (intervention), or a control group [22]. The intervention groups showed no significant differences from the first to the third visit. On the other hand, another study including 352 patients who finished active treatment with 6 months of access to a web-based self-management support system, showed an increase in strenuous exercise in the intervention group compared with the control group [37]. Another study found that moderate-intensity aerobic exercise for at least 150 minutes per week significantly increased in the intervention group compared with controls [32]. In the study by Beatty et al [23], 60 participants received either the 6-week intervention Cancer Coping Online (n=30) or the 6-week web-based attention control (n=30). The Cancer Coping Online participants had significantly higher physical functioning than the controls at 3 months of follow-up (d=−0.52; P=.02). The study that found no significant effect [18] was similar to a study that found a significant effect, with the only difference in follow-up time (6 months vs 12 months) [19].

Four studies reported dietary behaviors [18,19,32,37]. One study, including 59 patients with 12 weeks of access to a web-based self-management exercise and diet intervention support system, showed a greater improvement in overall diet quality in the intervention group (P=.001) [32]. Another study, including 352 patients who finished active treatment with 6 months of access to a web-based self-management support system, reported no significant changes in fruit and vegetable intake [37]. Both studies were classified as moderate quality. Two high-quality studies of 462 patients, who completed active treatment, found no significant changes in dietary behavior and vegetable intake in particular [18,19].

In this paper, we have systematically reviewed published RCTs to assess the effectiveness of existing interventions with DSMSTs on physical and psychosocial symptoms or other supportive care needs in adult patients with cancer. A total of 19 publications covering 17 unique studies were included in this review. The RCTs varied in terms of content, duration, and frequency. Nevertheless, we identified 5 elements that were common for the majority of the interventions: an assessment component, tailored symptom self-management support, an information section, a communication section, and a diary. Significant positive effects were observed on QOL in 6 studies, on anxiety in 1 study and depression in 2 studies, on symptom distress in 5 studies, on PA in 4 studies, on dietary behavior in 1 study, and on fatigue in 2 studies. In addition, significant negative effects were observed on anxiety and depression in 1 study. Other studies reported no significant effects on these outcomes (4 studies on QOL, 3 studies on anxiety, 5 studies on depression, 2 studies on symptom distress, 2 studies on PA, 3 studies on dietary behavior, and 3 studies on fatigue). Most interventions were web-based; 2 studies used mobile apps, and 1 study used a game as a DSMST. No relationship was found between the effectiveness of the studies and different components of the DSMSTs for patients with cancer, the quality of the study, and the impact on physical and psychosocial symptoms or other supportive care needs in adult patients with cancer.

Concerning the reported effects of DSMSTs on psychosocial and physical symptoms or other supportive care needs, comparable reviews also showed positive effects of DSMSTs. A systematic review [38] included 16 studies that examined internet-based support programs in patients with cancer. That review showed that internet-based support programs are effective in improving psychosocial and physical symptoms in patients with cancer [38]. Another systematic review included 17 studies that examined web-based mental health interventions in patients with chronic gastrointestinal conditions. That review showed that these interventions resulted in fewer somatic symptoms and improved QOL [39]. Moreover, another review suggested that DSMSTs could be useful for individuals during and after cancer treatment, especially in terms of information, follow-up planning, and management of side effects [14]. However, significant negative effects of DSMSTs were also observed in the reviewed studies. Several studies have reported no significant effects of DSMSTs on specific psychosocial and physical symptoms or other supportive care needs. Other factors may also have played a role in the large variation in observed outcomes. These include the different measurement instruments used within and between studies, different sample sizes, and different periods between the start of the intervention and the postintervention measurement. Future studies should preferably use uniform outcome measures and time intervals for the assessment of outcomes.

For some patients, having more knowledge about their condition might reduce their anxiety as a result of the development of realistic expectations of the future and preparedness for treatment-related side effects. On the contrary, information might also increase patients’ anxiety by drawing attention to their condition, unknown symptoms, or risks of treatment. In our review, one study [21] reported significantly lower anxiety and depression levels in the intervention group than in the control group, whereas another study [34] reported significantly higher anxiety and depression levels in the intervention group than in the control group. In the first study, patients had access to the intervention for 1 year and could use the system as much as they liked. In the second study, patients received a tablet computer 1 week before surgery and had to return the tablet 1 week postoperatively. However, the heterogeneity in content, frequency, and duration of the interventions included in our review precludes a definitive answer to the question on the effect of digital self-management support on anxiety and depression. An earlier review on web-based interventions for type 2 diabetes indicated that interventions of longer duration (>12 weeks) resulted in better outcomes. This may also be the case in patients with cancer [40]. However, further studies are needed to confirm this.

We considered the mode of delivery (how the intervention was delivered to the recipients) in the included studies and identified that the technological component was mainly a web-based approach. Over the past 10 years, the number of publications reporting on the use of DSMSTs in health care has increased. At the beginning of this period, studies focused on telehealth, whereas in the past 5 years, the majority of studies reported on the use of eHealth and mobile health in DSMSTs [7]. Mobile devices have emerged as an important tool for enhancing communication between patients and health care providers, for patient engagement in their health, for disease prevention, and for interventions that change health behavior [13,41]. Surprisingly, we found only 2 studies using mobile health apps, which is in contrast with the rapidly growing market of mobile health apps in general. This may imply that the introduction of new mobile health tools is much faster than its scientific appraisals [7]. To better understand why one device in a care program is more effective than others, adequately conducted studies that moderate the possible effects are needed.

We identified 5 elements that were common for the majority of the interventions: an assessment component, tailored symptom self-management support, an information section, a communication section, and a diary. These elements were used as single-component interventions or multicomponent interventions using different combinations of elements. Of the 16 web-based approach studies, 5 studies used all 5 elements in their DSMSTs [21,25,28,29,32,37] to increase self-management (support). Given the design of the studies, it is difficult to determine whether multicomponent DSMSTs are more effective than single-component DSMSTs and, in the case of a multicomponent DSMST, which particular component contributes most to a certain effect [13]. An earlier review of cancer survivors supported the benefit of an educational element, that is, cancer survivors who received sufficient information reported a better QOL [42]. Most reviews targeting cancer [11-15] highlighted that DSMSTs are mostly multicomponent and that there is a great deal of heterogeneity in the protocols and outcomes measured in cancer‐related DSMST studies. Future studies should be more structured to determine the role of individual intervention elements and should take the duration and frequency of interventions into account [13]. To further demonstrate the effects in patients, researchers should analyze and compare single-component and multicomponent DSMSTs separately.

Individuals have different preferences regarding information seeking, health care participation, and embracement of DSMSTs. Preferences of women and men might differ regarding health information seeking and support [43]. In addition, age might be of influence, as it might be more difficult to work with new technologies for the older adults [7]. Other factors that could influence the use of new technology are the educational level or the skills needed for using electronic devices [7]. The studies described in this review included patients with different disease or treatment phases. The effects of DSMSTs on physical and psychosocial symptoms or other supportive care needs might differ depending on the patients’ need for information and support, which may vary during the phases. Patients with cancer in the curative phase, for example, may need more information on how to cope with the late effects of surgery or chemotherapy, whereas patients in the palliative phase may want information about the self-management of pain and psychological distress [44]. Future reviews should focus more on comparing the effects of DSMSTs in different groups of patients, distinguished by treatment stage (curative or palliative) and tumor types.

Although our review was systematic and we took care to assure its quality, there are limitations to our study. One limitation of this review is that the studies included in the review were conducted predominantly with patients with breast cancer. In addition, several studies included in this review enrolled patients with mixed cancer populations. In some cases, the reported effects and evidence found in the included studies may apply more to one type of patients with cancer than to patients with other tumor types. The preferences and needs of patients with a specific tumor type may differ. Therefore, future studies should focus on specific tumor types. Another limitation is that although the average rating for methodology was good (13 studies were of high quality), the trials included in our review had several potential sources of bias and error. In particular, insufficient information regarding allocation concealment and the lack of blinding participants and personnel as well as outcome assessors might have biased the results. A third limitation is that we summarized different types of self-management support interventions for different types of patient groups and compared their benefits for patient self-management. This heterogeneity hampers firm conclusions regarding the effects on the studied outcomes. In addition, some studies comprised small sample sizes [24,26,32,34]. The absence of significant effects might be caused by a lack of power instead of the true ineffectiveness of the intervention. Due to the large variety of outcome measures, study characteristics, and components of DSMSTs, neither meta-analysis nor a comprehensive description of effect sizes was possible.

In conclusion, this review suggests that DSMSTs have a beneficial effect on the QOL. For effects on other patient outcomes (eg, anxiety and depression, symptom distress, PA, dietary behavior, and fatigue), the evidence is inconsistent and limited or no effect is suggested. A total of 5 elements that were common for the majority of the interventions included an assessment component, tailored symptom self-management support, an information section, a communication section, and a diary. We identified several lacunas in the available body of evidence regarding the effects of DSMSTs on patients with specific tumor types, patients with cancer in a specific treatment or disease stage, the design of technology, and especially the design of technology tailored to the patients’ needs. Future research should focus on specific tumor types, study different types of interventions separately, and assess the effects of specific interventions at different stages of disease progression.