To our knowledge, this is the first study to assess the impact of a home-based diet and exercise intervention on PF among elderly cancer survivors. In designing this study, we hypothesized that the intervention would achieve roughly half of the effect size observed by Morey et al39
in a more intensive, clinic-based study in a similar population. The forecasted effect size, a standard deviation (SD) of 0.23 in the change scores between the two arms, was close to that observed (SD = 0.19). This value is comparable to differences in PF scores observed between cancer patients and healthy age- and race-matched controls (SD = 0.22)7
and reported in other research as clinically significant (eg, abatement of a migraine headache, SD = 0.2142
). However, we did not achieve our targeted accrual and were unable to declare this difference as statistically significant. Therefore, we join a host of underpowered trials, although our grounds (insufficient recruitment) differ from other trials (inaccurate projection).43-45
With a sample size of 182 participants, we would have required an SD of 0.35 to be 80% powered to declare statistical significance. Nonetheless, the strong trends observed in Project LEAD suggest that interventions that potentially improve PF warrant further exploration because functional decline exerts a major impact on older patients' QOL and health care costs related to supportive assistance.9
Thus, the findings of this study provide a basis by which to estimate statistical power for future research and provide preliminary evidence that home-based interventions may work within this hard to reach, vulnerable, and rapidly expanding population. The minimal attrition noted throughout this 12-month study also suggests that such interventions would likely be well received, although our low dropout rate and significant dietary change could be influenced by the socially advantaged and highly motivated nature of our sample.
This home-based intervention, which addressed dietary change in several domains, resulted in significant pre- to postintervention improvements in diet quality; these changes were not statistically significant for individual nutrients or food groups, but they cumulatively contributed to an overall improved diet. Although the intervention did not produce changes in physical activity that achieved statistical significance, it is possible that the CHAMPS instrument, which categorizes weekly activity into blocks of time, lacked sensitivity to detect modest increases in exercise. For example, an increase of two 20-minute exercise sessions per week over baseline would not be detected using the CHAMPS. Thus, future intervention studies may include CHAMPS categories of physical activity but also collect continuous (minutes of physical activity) data as well. Of note, a statistically significant increase in self-efficacy for exercise was observed among members of the intervention arm, suggesting that improvements in exercise behavior are mediated through this construct.33
Changes in self-efficacy or stage of readiness were not observed with respect to diet, which is a finding that may be explained by high baseline levels suggesting a ceiling effect. Given that our dietary intervention included both additive (encouragement to consume diverse diets with more F&Vs and whole grains) and reductive (encouragement to limit consumption of fat, saturated fat, and cholesterol) strategies to improve diet quality, our attempt to measure self-efficacy and stage of readiness for consuming a globally improved diet may have failed. Future studies may be better served by assessing these measures on individual domain-specific factors (eg, dietary fat, F&V, whole grains, and so on), rather than assessing healthy diet as a whole.
At the 12-month interview, differences between study arms diminished for PF, diet quality, and physical activity. Recidivism with fat-restricted diets (a large component of diet quality) and exercise interventions is a historic problem,43-45
and potential solutions usually involve increases in intervention intensity and frequency of contact. Previous studies also suggest that changes in lifestyle behaviors require continuous adherence for roughly 6 months before they become in-grained,34
thus extending the intervention period to allow for continued support once individuals have adopted new behaviors may increase the likelihood of durable effects. That being said, it may be unrealistic to expect stability over time46-48
because gradual declines in PF, physical activity, and diet quality are notable in longitudinal studies of aging populations.49-51
Therefore, the design of future trials should consider interim measures to control for secular trends.
Data related to QOL and depression were consistent and suggest that both interventions improved psychosocial well-being. Although these improvements may be an artifact of our highly motivated, socially advantaged sample, these results reinforce the need for an attention control when psychosocial outcomes are considered.
Perhaps the most valuable findings of this intervention development study relate to issues of feasibility and the potential for conducting such research on a larger scale. Lessons learned appear in the following paragraphs and may provide useful information to researchers who plan to pursue similar studies.
Patients ascertained from cancer registries often do not have complete data to allow for patient contact, especially if physician permission for contact is a proviso for institutional review board approval. In our experience, 26% of patients were not able to be contacted because of missing physician information.
Physician permission to contact patients was denied for 16% of patients, with concern regarding the Health Insurance Portability and Accountability Act as the most frequently cited reason for nonparticipation, even though the protocol met Health Insurance Portability and Accountability Act standards.52
A response rate of approximately 34% was noted for this home-based diet and exercise intervention that targeted newly diagnosed elderly breast and prostate cancer survivors. Levels of interest were greater among whites and males and those who were younger and more proximal to diagnosis. However, most of those expressing an interest already reported regular exercise (54%), and 11% followed healthy diets. With recent findings indicating that only 24.9% of elderly cancer survivors are physically active,53
it is clear that our recruitment efforts yielded a biased sample. Thus, strategies are necessary to increase receptivity for diet and exercise interventions among cancer survivors who need and could benefit from such interventions. Oncologists could provide valuable assistance by supporting healthful lifestyle change.17
Future trials also need to budget adequate resources to accrue this population, which is acknowledged as hard to reach.54
Most (68%) newly diagnosed breast and prostate cancer survivors interested in participating in home-based diet and exercise interventions report no contraindications to unsupervised physical activity or an F&V–rich diet. Furthermore, the lack of differences noted between arms regarding adverse events suggests that, with appropriate screening, such interventions are safe.
High levels of agreement were noted between self-reported and clinically assessed BMIs, and significant correlations existed between self-reported walking items of the SF-36 and clinically assessed 6-minute walk tests. These findings provide evidence that telephone interviews performed in elderly populations yield valid information.
The low rate of attrition suggests that home-based lifestyle intervention studies are well-accepted among elderly cancer survivors. However, the recidivism observed in behavioral end points suggests a need for further research in developing interventions that produce durable effects.
Thus, Project LEAD provides valuable information. First, its process data can help inform other intervention trials that target older cancer survivors. Second, data suggest that home-based diet and exercise interventions can be safely delivered and improve lifestyle behaviors, which ultimately may improve PF. Given that Project LEAD is an initial foray into home-based lifestyle interventions among elderly cancer survivors, its approach holds promise and beckons for more research in this area, research aimed at producing durable improvements in behavior and function and that is adequately resourced to ensure accrual of this vulnerable and difficult to reach population.