A systematic review of the effects of remote feedback in home-based strategy physical activity interventions for cancer survivors


 Background: Studies have reported on positive health outcomes from physical activity (PA) for cancer survivorship. Systematic reviews, meta-analysis and trial studies have concluded that PA intervention improves quality of life and physical function, reduces the relapse of cancer and mitigates cancer mortality. Despite these important gains, cancer survivors’ PA level is still not satisfactory. Across all cancer types, home is the most preferred location of PA. This study aims to evaluate the effectiveness of remote feedback delivery strategy for home-based PA among cancer survivors. Methods: A systematic review was conducted by retrieving relevant articles published from the year 2009 to 2019 from the database aggregator EBSCOhost (databases used were Medline, CINAHL, SPORT Discuss and Cochrane trials). Results: Both frequent remote contact and direct remote contact are equally effective and have a larger effect than programmes with usual care, no intervention, participants placed in waiting list for intervention or participants given basic guidelines on PA. Direct remote contact reported strong evidence of moderate effect while frequent telephone contact reported conflicting findings of varied effect sizes. Conclusion: Opting for remote feedback strategies with supervised exercise provide an easier approach for the cancer survivors to keep up with exercise for a longer period of time and can be integrated in a cancer survivor’s lifestyle.


Page 3/15
This study aims to evaluate the existing literature on the effectiveness of remote feedback delivery strategy for home-based PA among cancer survivors. An examination of the evidence on this type of delivery strategy for cancer survivor enables better planning of PA intervention programme for this group of population. Home-based PA that integrates a properly designed remote feedback would effectuate the initiation and adherence of PA. Concurrently, when an appropriate remote method of supervising PA is chosen, a less cost-intensive, less time-consuming, less privacy-invading and more accessible programme can be offered as compared to supervised exercising.

Methods
Identi cation of the studies Relevant articles were identi ed by using EBSCOHost that retrieved articles from databases of CINAHL, Medline, SPORTDiscuss and Cochrane trials. The literature search was limited to articles published between 2009 and 2019. The principal search strategy was developed by Geraedts, Zijlstra, Bulstra, Stevens, & Zijlstra (2013) and was based on systematic reviews of the effectiveness of remote feedback delivery strategies in home-based PA among older adults [12]. The following search strategy and key terms were used tailored for this study objective: 1. Key terms #1: "homebased" OR home OR "home-based" 2. Key terms #2: remote OR stimulation OR coaching OR feedback 3. Key terms #3: monitoring OR telemonitoring OR telecommunication OR tele-communication OR telephone (NOT ''telephone survey'' OR ''telephone surveys'') OR physiotherapy OR ''physical therapy'' OR telerehabilitation OR tele-rehabilitation OR rehabilitation OR tele-exercise OR telecare OR tele-care OR teletraining OR teletraining OR telemedicine OR tele-medicine 4. Key terms #4: tness OR balance OR mobility OR exercise OR ''physical activity'' OR activity OR ''physical tness'' OR exercise OR ''motor activity'' OR ''psychomotor performance'' OR ''exercise movement techniques'' OR ''postural balance'' 5. Key terms #5: neoplasm OR cancer* OR "cancer survivor*" OR "cancer patient*" . Combination of key terms #2 or #3 (key terms 6): #2 OR #3 7. Combination key terms #1, #4, #5 and #6: #1 AND #4 AND #5 AND #6 Eligibility criteria Following the databases searched, duplicates were removed and the remaining references were screened based on their title and abstract by the rst author (MO) independently. Subsequently, the remaining articles were checked for relevancy on the research question. Any discrepancies in the selection of the articles were solved by consultation with the second author (NAMZ).
The following general inclusion criteria were formulated for the current study selection: (1) The study assesses PA intervention programmes that used remote feedback in the home It is de ned as any PA interventions which does not include physical contact or meeting except for effect measurements and explanation of or initiation into the program, aimed at enhancing their PA.
(2) The study design is a randomized controlled trial study (3) One of its outcome measurements is general PA behavior or physical capacity (4) The study participants are among cancer survivors Page 4/15 (5) Study with full text available and published (6) The article is written in English.
In this review, adherence was also analysed in studies that addressed adherence along with the eligibility criteria. Adherence is de ned as "the extent to which a person follows a prescribed exercise routine" [12].

Quality assessment
Quality assessment was conducted by using the PEDRO scale. For more information about the quality ratings see Additional 1.
A total of 11 items with a dichotomous answer of "yes' or "no" (1 and 0 points respectively) are being assess in the scale. Item 1 assesses for external validity or the eligibility criteria. Item 2 to 11 focus on the internal and statistical validity of the article.
Quality assessment was performed independently by the researcher (MO), and any uncertainties were solved by the second researcher (NAMZ). A score of above 5 from the cumulative of item 2 to 11 was considered adequate trial quality and above 6 is considered high quality [13,14]. Only studies above than score 5 from the cumulative of item 2 to 11 will be included in the review. Five levels of evidence were classi ed from the quality assessment results as described below: 1. Strong evidence: generally ≥75% of the studies reported consistent ndings from multiple high rated quality studies.
2. Moderate evidence: generally ≥75% of the studies reported consistent ndings in one high quality study and one or more low-quality studies, or generally consistent ndings in multiple low-quality studies.
3. Limited evidence: only one study in one high-or low-quality. 4. Con icting evidence: <75% of studies reported consistent ndings 5. No evidence: no randomized controlled trials (RCTs) or non-RCTs.

Data analysis
Data extraction was performed from the included articles to perform the effectiveness analysis. Each outcome from the studies was calculated by using the standard mean difference (SMD) with a corresponding 95% con dence interval. The formula used for standard mean difference (SMD) is described below: Following that, the values were interpreted as either a small effect (0.2-0.4), moderate effect (0.5-0.7) or large effect (0.8) [13]. Data summarization was performed using a qualitative best-evidence synthesis [14].

Description of studies
The full study selection following PRISMA ow 2009 diagram is shown in Figure 1. Records identi ed using the database search from EBSCOHOST of MEDLINE, CINAHL, SPORT Discuss and Cochrane trials were 413 articles. After removing duplicates, the number of articles that remained were 350. Following the screening process based on title and abstracts, 32 articles were included. The nal number articles included was 9 after removing those that were not experimental studies (n=10), had no protocol development (n=6), did not involve remote feedback for PA (n=5), did not report PA outcome measures (n=1) and was a non-randomized controlled trial study (n=1).

Types of analyses
Six studies reported both effectiveness and adherence analysis [15][16][17][18][19][20][21][22] while three studies did not report any adherence analysis [23][24][25]. Apart from the outcome measurements is PA level or PA capacity, there is heterogeneity of the secondary outcome or other outcome measures in the included studies as shown in Table 1.

Description of study samples
The range of the sample sizes was between 17 and 410 and the range of participant age was between 31 years and 68 years.
The overall mean age of participants in all included studies was 53.8 years. The types of cancer group included in the studies were varied. However, breast cancer and colorectal cancer (CRC) were among the cancer groups most frequently studied. Two studies focussed on young adult cancer [18, 19,23] and CRC [20,25] and one study each on breast cancer [24], hematologic cancer [16,17], combination of breast cancer and CRC [21], combination of breast, prostate and CRC [22] and all types of cancer [15] ( Table 1).

Quality of studies
Some studies lacked information to indicate randomisation, allocation concealment, blinding, and ancillary analysis.
According to the PEDro scores calculated, there were three moderate quality studies [23][24][25] and six high quality studies [15][16][17][18][19][20][21][22]. None of these studies were rated as poor quality ( Table 1). The range of PEDro scores from the included studies were between 6 and 9 of a maximum score of 11. The mean of the total score of the quality of all included studies was 7. In subanalysis, scores range for internal validity was between 5 and 8 (maximum score 10) with a mean score of 6. For more information about the quality ratings see Supplement 1.

Effectiveness of remote feedback strategies
Details on the effectiveness of remote feedback strategies are described in Table 1. The remote feedback strategies in the included studies were varied. Therefore, the remote feedback strategies were classi ed into 3 categories (direct remote contact, frequent telephone contact and non-frequent telephone contact) [12] as described below: (1) Frequent telephone contact: more than one phone call per month, (2) Non-frequent telephone contact: less than one phone call per month and (3) Direct remote contact: studies using direct remote strategies during PA other than telephone contact Frequent telephone contact The frequent telephone contact category has six studies with three studies [23][24][25] rated as moderate quality and three studies [16,17,20,21] rated as high quality. Effect size was unable to be calculated for three of the studies. Findings in the remaining three studies with measurable effect size showed con icting evidence as less than 75% of them reported inconsistent ndings in the PA outcomes. Two of the moderate quality studies reported improvement in both intervention group (IG) and control group (CG) for PA outcome. Three of the studies were able to calculate for standard mean difference [16,17,20,21]. The effect size of those three studies ranged between 0.42 to 1.27 resulting in the average effect size of 0.75. Each of the studies with measurable effect size was classi ed as low effect [21], moderate effect [20] and strong effect [16,17]. These ndings illustrate that frequent telephone contact showed a comparable or signi cantly better results on PA measures for the IG compared to the CG.

Direct remote contact
The three studies that remotely contacted participants directly were rated as high-quality studies [15,18,19,22]. Strong evidence of moderate effect size were in two of the high quality studies. All studies were able to calculate for standard mean difference with the effect size range of between 0.13 and 0.76, resulting in the average effect size of 0.49 or moderate effect. These ndings illustrated that frequent direct contact showed a comparable or signi cantly better results on PA measures between the IG and the CG.
Adherence in remote feedback strategies An overview of the adherence results in remote feedback strategies is described in Table 1. Four of the included interventions reported adherence results [18][19][20][21][22]. The adherence in IG varied widely from 16.4% [20] to 90% [21]. Two of the six interventions from frequent telephone contact provided adherence results, while two of the three interventions from direct remote contact provided su cient information in the articles for adherence analysis.

Discussion
This systematic review presents an overview of literature pertaining to remote feedback delivery strategies in home-based PA interventions for cancer survivors. Geraedts et al., (2013) have conducted a similar systematic review but on a different study population which is older adults [12]. On top of effectiveness analysis, additional analysis was conducted here including adherence analysis. Both frequent telephone contact and direct remote contact were proven to have bene ts at some extent in effectiveness but with varied strength of evidence.
Findings from this systematic review show that both frequent remote contact and direct remote contact are equally effective or has a larger effect than usual care, usual care, without any intervention, placed in waiting list for intervention or given basic guidelines on PA. Direct remote contact showed strong evidence of moderate effect size, while frequent telephone contact has shown con icting evidence of varied effect size. However, the average effect size on each category favoured towards frequent telephone contact (0.79) as compared to the direct remote contact (0.49).
Studies using the frequent telephone contact strategy, most reported contact the participants once in a week with at least fortnightly contact and a few reported of biweekly contact with the participants. Frequent remote contact has a greater effect size value as compared to remote direct contact. However only three out of six studies using frequent remote contact provide available data of PA measures to determine the standard mean difference. For direct remote contact, all the three high rated quality studies calculated the standard mean difference and the effect size for PA measures could be determined. In addition, there is heterogeneity in the frequent telephone contact method in which two of the studies did integrate other strategies of remote feedback such as newsletters [24] and text messages [25]. Unfortunately, none of the aforementioned these studies were able to calculate for effect size. Furthermore, wide heterogeneity is seen particularly in studies using frequent telephone contact. The studies varied widely in terms of the target cancer groups, duration of intervention and outcome measures. This heterogeneity is probably the cause of the con icting results for effectiveness in the frequent contact group where some of the reported variables produced a mixed of negative and positive effect sizes although the overall effect size was large.
Meanwhile, for studies using direct remote contact, all the studies reported the variables that enabled the effect size to be calculated. It seems that direct remote contact is equally effective or has a larger effect than usual care or if there is no intervention consisting of PA measures. The strategies used for direct remote contact differ between each study. James et al.
(2015) used pen-and-paper mailed [15], Valle et al. (2013) used a combination of Facebook and website [19] and Forbes, Blanchar, Mummer, & Courneya (2015) decided to employ a combination of websites and emails [22]. With the recent technology advancement where people can get connected everywhere and at any time, this shows that direct remote contact is a promising alternative to get in touch with participants remotely while monitoring their PA at home. Yet, more research is needed to look at various modes of direct remote contact to provide a strong evidence on its effectiveness.
Unfortunately, in this review none of the RCTs has been classi ed under non-frequent telephone contact. Therefore, comparison cannot be conducted between this method of remote feedback delivery strategy with either frequent telephone contact or direct remote contact.
More than half of the included articles that employ remote feedback strategies were developed based on speci c theoretical frameworks. These consist of three studies that used frequent telephone contact [21,23,24] and two studies that used direct remote contact [15,18,19]. All of these studies are based on Social Cognitive Theory except for one study used a combination of Social Cognitive Theory and Transtheoretical model [15]. Generally, theory-based interventions have shown a promising positive effect size when remote feedback strategies are used.
In this review, adherence results were extracted as part of the analysis. However, they were unsystematically inventoried in the studies. Adherence results for studies using frequent telephone contact have both the highest and lowest adherence ndings.
This illustrates the inconsistency of adhering to frequent telephone contact as a remote feedback delivery strategy. Based on the reported adherence information, consistent acceptable adherence to interventions using direct remote contact is seen in IG, varying between 43.7% and 58%. From these ndings, direct remote contact for home-based exercise is determined to be a promising alternative to supervised onsite exercise. Although this review does not have any study categorised under nonfrequent telephone contact, ndings from a review of older persons using this method yielded a good adherence result [12].
However, it could not be concluded in this review as this type of remote feedback delivery strategy was not conducted among cancer survivors.
Finally, limitations in this review should be mentioned. Following the exact words used in the search strategy, studies that used remote feedback delivery strategies but were not using home-based or remote feedback were not included in the reference list.
One of the selection criteria was that the studies must be reported in English language. Potential bias on language selection might limit ndings from our articles. A few of the studies were also unable to provide su cient data to allow us to calculate for standard mean difference particularly in the case of frequent telephone contact as a remote feedback delivery method. Therefore, the overall effect size might be not accurate for frequent telephone contact and the next method of best evidence synthesis is used since statistical evidence could not be provided.

Implications To Practice
Although in this review frequent telephone contact and direct remote contact gave favourable ndings, researchers need to be cautious in employing such remote feedback strategies in their intervention programme. Frequent telephone contact might be cost intensive in a large scale intervention study while choosing an appropriate direct remote contact might be equally as effective as frequent telephone contact in addition to being less cost-intensive, less time-consuming and less privacy-invading as well as providing a wider sample coverage. Secondly, in a community-based or centre-based setting, direct remote contact coupled with a trainer can be an alternative for supervised exercising.

Page 8/15
Overall, home-based exercise with remote feedback strategies provide additional bene ts by simplifying the logistic aspect and can be integrated easily with a cancer survivor's lifestyle. However due to the various health condition of cancer survivors, supervised exercising in addition to remote feedback strategies provide a more effective approach for cancer survivors to keep up with exercise for a longer period of time and enjoy the associated health bene ts.
Abbreviations CG: control group; IG: intervention group; PA: physical activity; SMD: standard mean difference; WHO: World Health Organization Declarations Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
All data generated or analysed during this study are included in this published article.

Competing interests
The authors declare that they have no competing interests.

Funding
The authors received no nancial support for the research and authorship of this article.
Authors' contributions MO performed article review, data extraction and synthesis. NAMZ provided overview and consultation during the whole process. All authors contributed to writing the manuscript. All authors read and approved the nal manuscript.