All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Participants were 37 cancer survivors who had previously participated in a survey about health behavior change interventions and delivery modalities and indicated that they were willing to be contacted for participation in future studies. Participants had to have a diagnosis of either locoregional breast cancer (stages 0 to IIIA), colon cancer (stages I and II), endometrial cancer (stages I to IIIa), or prostate cancer (stages I and II) and no history of any other cancers. Participants were required to be at least 3 months postsurgery (if applicable), over the age of 18 years, have a BMI ≥25, have access to high-speed Internet and a telephone, and live in the Houston area. Survivors were excluded if they had a medical condition that prevented them from engaging in an unsupervised exercise program or low-fat diet high in fruits and vegetables. Informed consent was obtained from all individual participants included in the study. In-person assessments were completed at baseline and 6 months for all measures.

This study was approved by MD Anderson’s Institutional Review Board and was registered at ClinicalTrials.gov [NCT01311856]. Following the baseline assessment, participants were randomized at a ratio of 2:1 to either the Internet-based weight loss intervention or the telephone-based version, respectively. A 2:1 ratio was used because we hypothesized that outcomes in the Internet condition would be smaller. We used a form of adaptive randomization called minimization, which is similar to stratification in that participant characteristics are used to assign them to the treatment conditions [32,33]. All participants received resistance bands and pedometers. Participants in the telephone intervention received print materials about exercise and diet and telephone counseling calls (3 weekly, 2 semiweekly, 4 monthly; 15 to 30 minutes in length) and customized mailed progress reports every 6 weeks to encourage adherence to diet and exercise recommendations. Materials were based on the Reach Out to Enhance Wellness (RENEW) intervention [16]. Participants in the Internet-based intervention had access to the same content online by logging onto www.walkingspree.com/login/healthymoves with a personalized username and password. Participants in the Internet arm were also invited to participate in a discussion forum facilitated by intervention staff, had the opportunity to email questions directly to the intervention staff, and received customized progress reports every 6 weeks by email.

The goals for both groups were to do 15 minutes of strength exercise every other day, ≥30 minutes of walking or other moderate-intensity exercise on 5 or more days of the week, and consume a diet with 7 (for women) or 9 (for men) servings of fruits and vegetables per day and <7% of calories from saturated fat. Participants in both groups were also provided with caloric recommendations to facilitate a weight loss of 1 to 2 pounds per week (a loss of 5% body weight was used as a goal over the course of the 6-month study period) and fat gram/calorie counters or access to appropriate websites to monitor intake. Participants received 2 $25 gift cards as compensation; 1 after completing the baseline assessment and the other after completing the 6-month assessment.

Two-sample and paired t tests and Fisher exact tests with a 2-sided alpha of .05 were used to (1) compare the 2 intervention groups on a number of demographic variables; (2) compare the difference scores from baseline to 6 months on diet, physical activity, physical performance, and body composition between the 2 intervention groups; (3) assess within-group changes from baseline to 6 months on the aforementioned outcome variables; and (4) compare attrition rates between intervention groups for each outcome variable. We define attrition here as any participant who completed baseline measures but stopped participating at some point following baseline (eg, the participant dropped out of the study and no further data were collected). Additionally, Cohen d was calculated for within-group differences between baseline and 6 months.

This pilot RCT provides some of the first data directly comparing telephone- and Internet-delivered weight loss interventions among a sample of cancer survivors. Results suggest that the engagement was far greater with telephone intervention and consequently yielded larger improvements in several measures of body composition (especially waist circumference, which was highly significant), diet, physical activity, and physical fitness. Although outcomes generally favored the telephone group, participants with a higher percentage of body fat were more likely to drop out of this intervention group, indicating that the Internet intervention may be more acceptable for people with a high percentage of body fat. Previous research is mixed with regard to why participants drop out of weight loss interventions, but some research has found that for in-person interventions, weight or shape concerns may increase the likelihood of attrition [34].

Although our modest sample size and lack of statistical power hampered our ability to detect significant differences, several differences are worth noting as the effect sizes are clinically meaningful, including the percentage of body fat, fruits and vegetables consumed, moderate/vigorous physical activity (measured via accelerometer), and 6-minute walk test (found in Figures 2,4, and 5, respectively). Changes in these variables for participants in the telephone group were in the hypothesized direction, while participants in the Internet group showed changes in the opposite direction. In a larger sample, these differences may have been more pronounced. These findings are consistent with a recent review of telephone- and Web-based weight management interventions for cancer survivors which suggests telephone interventions may be more effective than Web-based approaches [15].

In terms of within-group change, the telephone group had more outcomes related to fitness and weight loss that changed over the 6 months than the Internet group. These included weight, waist circumference, and 8-foot up-and-go time. In terms of change in weight, participants in the telephone group experienced a 5.6% weight loss which is clinically meaningful [35]. Participants in this group also experienced a decrease of 5 cm in waist circumference, decreasing from 97.2 cm to 92.4 cm. Epidemiological research suggests an increased risk of all-cause mortality among individuals whose waist girth falls within the range of 95 to 100 cm as compared to those whose waists measure 90 to 95 cm (especially among women) [36]. Interestingly, the Internet group did have 2 measures of physical functioning change including the 30-second bicep curl and sit-to-stand, which suggest some advancements in strength training. This group also showed a significant increase in percentage of body fat; however, results indicating changes in physical functioning may be an artifact of multiple comparisons as there were no meaningful differences between interventions other than modality.

By examining the engagement data we can indirectly assess participant perceptions of usefulness or enjoyment. Engagement was assessed in the telephone group via the number of phone sessions out of 9 that they completed. In the Internet group, it was assessed as the number of days that they logged on during the intervention out of a possible 160 days. This finding is consistent with the reviews and meta-analyses of the previous literature [17,18,37], which suggest that more personal contact with participants leads to better improvements in outcomes. Moreover, these reviews reported that interventions with at least one in-person interaction resulted in greater engagement and better outcomes. In a recent systematic review of weight loss intervention for cancer survivors [38], the authors note that interventions that combined technology-based modalities (such as telephone) with in-person counseling were more effective than those using only one modality.

Engagement in Internet-delivered interventions is an ongoing area of research. One recent study found that an Internet-delivered intervention for cancer-related distress among survivors suggests that engagement tends to be higher for women, for participants who underwent chemotherapy, and when participants are recruited online [39]. The authors also note that the social networking component increased overall engagement but may have interfered with other intervention components. In a separate weight loss study using a Web-based intervention, researchers found no difference between an information-based website and 2 supportive ones—one that provided feedback, social support, and self-monitoring and another that provided the same features plus personalized planning [40]. Despite the lack of significant differences between websites in terms of weight loss, use of the supportive websites was higher compared to the informative website, suggesting that greater engagement may not lead to greater weight loss. It should be noted that completing 9 telephone counseling sessions may not be equivalent to logging on to the website every day. Moreover, we did not have any measure of the pattern of log-ins over time or what the participant did while they were logged on, limiting our ability to infer how much of the intervention material to which they were exposed.

We compared percent completion of online surveys across the 2 groups, and while differences were not statistically significant, studies with larger sample sizes may find significant differences. In order to address this issue, standardized measures of engagement should be developed to compare across Web-based and non–Web-based interventions. One possible measure may be length of time exposed to intervention materials (eg, length of telephone sessions in minutes and number of minutes participants spent logged in to the website). Researchers should continue to evaluate different measures in order to identify the most accurate measure of adherence.

This study has several strengths worth noting. First, it is one of the first studies to directly compare a telephone-delivered intervention and an Internet-delivered intervention in a sample of cancer survivors. Second, several objective measures of body composition, physical activity, and physical performance were used to capture changes in important markers of health that may have occurred during the intervention. Last, the intervention that was delivered is easily disseminable and requires fewer resources compared to interventions that use supervised exercise sessions.

Although this pilot study had many strengths, there also were weaknesses. First is the small sample size. Few of the changes from pre- to postintervention were statistically significant; however, many of the between-group differences in outcomes, while not statististically significant in this study, would have likely been significant had the sample size been larger. Additionally, attrition for this study was fairly high with about 35% of participants dropping out before completing the study. Given the issues we encountered with attrition, future studies examining Web-based interventions for weight loss should account for potentially high attrition rates in the Web-based group by having proportionally more participants for this condition relative to other conditions. In the telephone group, participants with a higher percentage of body fat were more likely to drop out, and because of the small sample size, we were unable to control for this in our analyses. Finally, because many outcome variables were measured, many statistical comparisons were used, which could have increased the type I error rate; however, given that (1) this was a pilot study, (2) the goal of the pilot study was to identify relationships for future study, and (3) all comparisons were planned before the intervention was delivered, we felt that it was unnecessary to adjust for multiple comparisons.

The results of this pilot study are compelling and provide direction for future studies. Specifically, future studies that compare telephone- and Internet-delivered interventions would benefit from techniques to enhance adherence and examine cost differences. It may prove beneficial to augment current weight loss interventions in health care settings with personalized intervention components. Research suggests that interest in technology-based interventions is influenced by the survivors’ current technology use, their age, and their current lifestyle patterns (eg, eating and physical activity behaviors) [41]. In fact, a program using both telephone and Web components may be able to maximize reach and engagement. Future studies should also focus on how to get older participants to engage more with technology so as to enhance Internet-based interventions. Over time, as younger participants who are more comfortable with technology age, there may be a shift in preference of intervention modality toward Internet-based or other technology-based interventions. A recent study involving breast cancer survivors showed moderate improvements in weight (2% weight loss), fruits/vegetable consumption (+1.5 servings/day), and physical activity (+5.75 metabolic equivalent of task hours per week) in an intervention using a multimodal mHealth approach [42]. Several outcome measures showed promise in terms of 6-month change including percentage of body fat, waist circumference, fruits and vegetables consumed, moderate/vigorous physical activity (measured via self-report), and 6-minute walk test. Future studies should focus on these outcomes.

Finally, studies should also determine the most effective intervention components and how to best combine these in order to create the most robust intervention strategy. As Hoedjes [38] notes, several promising theoretical components such as goal setting, action planning and social support may be effective for weight loss interventions for cancer survivors; however, optimizing the modality for delivery may be just as important. Future studies should use the multiphase optimization strategy to determine the most effective components [43].