We first searched PubMed, EMBASE, Google Scholar, CINAHL, and PsycINFO in July 2013 for prospective RCTs evaluating telehealth in cancer care regarding pain, depression, and quality of life. The search was updated in January 2015.The keywords were as follows: “neoplasms [MeSH]”, “cancer” and “Remote Consultation [Mesh]”, “mHealth”, “connected health”, “text messaging”, “telemedicine”, “telehealth”, “ehealth”, “telephone therapy”, “teleconsultation”, “mobile technology”, “telecare”, “Internet”, “digital health”, “mobile phone*”, “smartphone”, “apps”, and “mobile application”.

We included RCTs that met all of the following criteria: reported the effect of telehealth on pain, depression, or quality of life in cancer patients. If data were duplicated or shared in more than one study, the last published or more comprehensive study was included in the analysis.

Based on the pre-determined selection criteria, 2 authors (JW, SA) independently selected all trials retrieved from the databases and bibliographies. Disagreements between evaluators were resolved by discussion.

The methodological quality of included studies was assessed by the Cochrane Collaboration’s risk-of-bias tool [11], a commonly used tool to report the risk of bias in individual studies included in systematic reviews. The tool assesses several internal validity domains, which include sequence generation, allocation concealment, blinding of study participants, personnel and outcome assessors, incomplete outcome data, selective reporting, and other sources of bias. We classified each of these domains as being at high, low, or unclear risk of bias. Data were entered into Review Manager 5.3, and a risk of bias graph was generated for all included studies.

Figure 1 depicts a flow diagram of how we identified relevant clinical trials included in this review. Using the above-mentioned keywords, a total of 4929 articles were identified from the literature search of five databases, that is, PubMed, EMBASE, Google Scholar, CINAHL, and PsycINFO. After excluding 297 duplicated articles, 2 authors independently reviewed and excluded an additional 4561 articles that did not satisfy the pre-determined selection criteria based on each article’s title and abstract. We reviewed the full texts of the remaining 71 articles and excluded 51 articles because of the following reasons: identical trials with the same population (n=7), nonrandomized studies (n=8), trials not related to the subject (intervention/outcome) of this study (n=32), and trials reporting only the study protocol (n=4). A total of 20 trials were included in the final analysis [12-31]. Since the studies included in this review are RCTs with comparator groups, we report only between-group effect estimates. Pre- and post measures within groups were not considered.

Table 1 shows the general characteristics of the 20 trials. Eight (40%) of the 20 studies focused on improving quality of life [16,18,19,22,24,26,28,31], another nine (45%, 9/20) on improving depression outcome (Sherman et al actually focused on psychological well-being) [12,15,17,20,21,25,27,29,30], two (10%, 2/20) on improving pain control [13,14], and one (5%, 1/20) has both pain intensity and depression as main outcomes [23]. The Functional Assessment of Cancer Therapy (FACT) was the most (62.5 of all studies with HR-QOL as main outcome) commonly used measure of quality of life. The Center for Epidemiologic Studies Depression Scale (CES-D) and the Hospital Anxiety and Depression Scale (HADS) were the most commonly used measure for depression and the Brief Pain Inventory most commonly (66.7% of all studies with pain intensity as main outcome) used to evaluate pain outcomes. The studies were published over a period of 9 years from 2006-2014.

The majority (13/20, 65%) of the studies were conducted in the United States. The other countries represented include Australia (2/20, 10%), South Korea (2/20, 10%), Sweden (1/20, 5%), Finland (1/20, 5%), and Norway (1/20, 5%). The sample size in each of the studies ranged from 25-571 for a total of number 3789 subjects in all. The median follow-up time was 4 months with a range of 1 week to 18 months. Thirteen (65%) of the 20 trials, were conducted among patients with breast cancer [12,16-18,21,22,25-31], followed by four trials in patients with any solid cancer [13-15,23], and one trial each with focus on cervical [24], colorectal [19], and colorectal/prostate cancers [20].

Many of the included studies (14/20, 70%) were telephone-based interventions, although two of them were used in conjunction Web-based systems. Most of these telephone-based interventions (12/14, 85.7%) involved a professional interventionist (nurses, psychologists, or counselors) trained to provide counseling, while the remaining two studies [18,25] were delivered by peer counselors who are cancer survivors. Additionally, only one of these telephone-based studies utilized automated voice response [23], which was actually used in conjunction with life-support personnel. Five of the studies [15,27,29-31] used Web-based delivery systems for their interventions, and one study [28] utilized store-and-forward video-recorded sessions to deliver their intervention. The duration and frequency of the interventions varied and so also the total intervention time with a median of 12 weeks and range of 1 week to 12 months. Table 2 summarizes the main results from each of the studies showing effects of the intervention on primary outcomes.

Figure 2 depicts the methodological quality of the studies included in this review. Most of the studies provided information about the method of generation of random sequence, while two studies [14,20] applied inappropriate methods of random sequence generation. Only a few studies (25%, 5/20) provided information about allocation concealment. Similarly, only a few studies (25%, 5/20) [15,17,23,29,31] had low risk of bias on the blinding of subjects and study personnel domain. Seven studies [13,15,18,21,22,24,27] were judged to have a risk of bias on the incomplete outcome reporting because of imbalance in dropout rates by group or insufficient accounting of all study participants.

Kim et al [14] compared the efficacy of pain education alone (control arm) and pain education plus telemonitoring (experimental arm) on pain and depression in a total of 108 patients with advanced solid tumors. In their trial, nursing specialists provided video-assisted educational material in both arms and daily telemonitoring for the first week in the experimental arm. There was significant improvement in pain and depression outcomes comparing baseline and final outcome in all study participants. They also reported significant reductions in number of intervention subjects with pain intensity scores ≥4 compared with control group (35%-19%, P=.02). Although average pain score over the past 24 hours (-1.2 vs -1.9) and worst pain scores (-0.7 vs -1.9) decreased compared to control group, these were not significant. Similarly, Harrison et al [19] evaluated the effectiveness of a nurse-delivered telephone supportive intervention (the “CONNECT” intervention) compared with usual care in 75 colorectal cancer patients. The CONNECT intervention consisted of five calls from a specialist nurse in the 6 months after initial discharge from the hospital [32]. They also found time-dependent improvement in HR-QOL within each arm but failed to reach a statistical significance comparing intervention and control groups.

Livingston et al [20] enrolled 571 newly diagnosed male CRC (n=182) and prostate (n=389) cancer patients and randomized them into three arms: two intervention arms and a passive referral arm. In the active referral arms, the specialist actively referred men to a Cancer Helpline. In Active Referral-4, patients received calls from the Helpline within 1 week of diagnosis, at 6 weeks, 3 months, and 6 months post diagnosis. In the Active Referral-1 arm, patients received only one call within 1 week of diagnosis. In the control arm, Passive Referral, patients were referred to contact the Helpline at their own initiative. The telephone helplines were developed by many cancer organizations in Australia to provide information tailored to the cancer patient’s needs, support, and referral to supportive service [33]. The study included only male patients based on prior work that suggested that men were less likely to utilize supportive services [33,34]. However, they found no psychological impact of the telephone-based intervention; mean changes over time in cancer-specific depression outcomes were similar between study arms.

In 2010, Kroenke et al reported the results of the Indiana Cancer Pain and Depression (INCPAD) trial [23]. In this trial, 202 patients were randomly assigned to receive the intervention and 203 to receive usual care. Patients in the intervention group received centralized telecare management by a nurse-physician specialist team coupled with automated home-based symptom monitoring by interactive voice recording or online. They reported that the intervention resulted in improved pain and depression outcomes in cancer patients assigned to receive the intervention. The standardized effect size for between group differences at 3 and 12 months was 0.67 (95% CI 0.33-1.02) and 0.39 (95% CI 0.01-0.77) for pain, and 0.42 (95% CI 0.16-0.69) and 0.41 (95% CI 0.08-0.72) for depression.

This systematic review of randomized controlled trials evaluating the effect of telehealth interventions on pain, depression, and HR-QOL outcomes in cancer care included 20 studies published over a 9-year period from 2006-2014. From this review, we make a number of observations. First, the application of telehealth in cancer care is in early stages, and all the studies were conducted in high-income countries. Second, the studies were largely heterogeneous in design and outcome assessments, making it difficult to pool effects in a meta-analysis. Third, the interventions are diverse in terms of type, content, intervention times, follow-up periods, and outcome measures.

Two of the three studies included in this review that focused on pain control reported a positive effect of telehealth on improving outcomes [14,23]. However, the study by Rustoen et al attributed inadequate dose of the psychoeducational intervention as one of the probable reasons for lack of efficacy. This is in contrast to the other two studies that relied heavily on collaborative care management between patients, their caregivers, and health care providers with extensive patient education. The INCPAD trial also included automated symptom monitoring as part of the intervention. Previous systematic reviews have identified patient education as a key component in improving cancer pain management. Lovell et al proposed four core principles that should guide the basis for patient education to successfully improve cancer pain management. These include the principles that education should be (1) patient-centered, (2) be an integral component of the patient-provider relationship, (3) aimed at patient empowerment for self-management, and (4) incorporated as part of ongoing care to counsel and support patients in the context of the severity of their pain, their needs and self-management plans. Therefore, incorporating collaborative patient-centered psychoeducational strategies in the design of technology-based interventions for pain management could improve efficacy of technology-based pain management interventions.

In contrast to the pain-focused studies, three of the eight studies evaluating the effect of telehealth on HR-QOL demonstrated improved outcomes. A recent review by Dickson et al evaluating the use of technology-based interventions for cancer follow-up surmised that the interventions did not decrease HR-QOL [35]. Similarly, four of the ten studies evaluating the effect of telehealth interventions on depression demonstrated significant improvement in depression outcomes. The SMART oncology trial, another collaborative care management approach that was delivered by a care manager under the supervision of a psychiatrist, demonstrated improvement in depression outcomes [36]. Also, a meta-analysis evaluating the effect of various interventions on depression in cancer patients showed that compared with controls, psychotherapy significantly improved outcomes and cognitive behavioral therapy was particularly associated with better outcomes [37].

It is noteworthy that the majority of interventions reported in this review were telephone-based. This is unsurprising because telephone systems are one of the oldest and most reliable information technologies available today, which makes them a very popular communication tool across different generations. They also enable personal live interactions between patients and providers, which can enhance the patient’s sense of being supported. While they are also very cheap, the cost of maintaining a professional to deliver care coupled with the fact that treatment effects could be provider-dependent may hinder scalability. While there are no doubts that the current dominance of telephone-based interventions will continue, current trends suggest that mobile phones will be the primary medium of delivery. This is evidenced by the near global ubiquity of cellular coverage and the increasing affordability, portability, and ease of use of smartphones [38]. Current estimates suggest that about 56% of US adults own a smartphone, and we envisage that this upward trend will continue [39]. While the telephone-based interventions in this review are largely voice communications, text messaging now appears to be a dominant function that is being used to engage patients across multiple disease groups.

Although not specific to connected health-related studies, previous studies have highlighted similar challenge [40]. Okuyama et al in their review of clinical trials evaluating psychosocial telephone interventions in patients with cancers and survivors also reported a similar finding that the majority of the studies reviewed lacked a standardization of outcome assessments and did not adhere adequately to reporting according to CONSORT guidelines. To standardize the reporting of technology-based interventions, the CONSORT-EHEALTH Group developed checklists to provide useful guidelines in reporting technology-based trials [41]. The time is ripe to capitalize on the current optimism of the potential of telehealth to transform care delivery. To realize this goal, we cannot overemphasize the need to design high-quality trials to comprehensively establish evidence of the effectiveness of telehealth-related studies.

This study is not without limitations. The fact that we included only studies reported in English could have led to the exclusion of relevant studies, but we do not believe this will significantly impact our findings. We limited our search to five databases and also did not search the gray literature to find relevant studies nor did we include non-randomized, retrospective studies. We believe that evidence from prospective randomized trials will be sufficient to demonstrate effectiveness. In addition, there was a heterogeneity of outcomes, outcomes assessment measures, and comparators, which makes it difficult to estimate overall effects.

This is one of the first studies seeking to evaluate the effect of telehealth on pain, depression, and quality of life outcomes in patients at different stages of their cancer experience. While the studies evaluating cancer pain outcomes proved to be effective, the same could not be reported for those evaluating depression and quality of life outcomes. In total, our findings suggest that the application of telehealth in cancer care is still at a very early stage and is mostly utilized in developed nations. Evidence of its effectiveness demonstrates promise of improving pain, depression, and HR-QOL-related outcomes in cancer patients. There is a need to invest resources into developing rigorous larger-sized clinical trials, standardize outcome assessments, and improve reporting of clinical trials to demonstrate the effect of telehealth and realize the potential of transforming care delivery.