Consider the Source: Examining Attrition Rates, Response Rates, and Preliminary Effects of eHealth Mindfulness Messages and Delivery Framing in a Randomized Trial with Young Adult Cancer Survivors

Abstract

Purpose: Young adults with cancer often experience stress, depression, and anxiety. Mindfulness meditation is an effective intervention for these outcomes, and maintenance support may be needed for long-term improvements. eHealth technologies provide a promising delivery strategy for maintenance interventions.

Methods: Following an 8-week mindfulness-based stress reduction (MBSR) course, 62 young adult cancer survivors were randomized to 8 weeks of instructor-framed messages, peer-framed messages, or no messages. On average, participants were 33.6 years old. The majority of participants were college-educated Caucasian females. We examined attrition rates between participants who received messages and those who did not, and compared response rates from different perceived sources. In addition, we evaluated the preliminary effects of eHealth support on mindfulness and associated outcomes.

Results: No significant differences in attrition or message response rates across groups were observed. Repeated measures models revealed significant group by time interactions on perceived stress, anxiety, and depression. There were no differences between the groups that received eHealth messages and the group that did not. There was a significant difference in anxiety symptoms from post-MBSR to post-messaging between messaging groups. Individuals who received instructor-framed messages reported increased symptoms of anxiety over time.

Conclusion: Attrition and response rates did not differ across groups, suggesting that eHealth may be a feasible strategy for providing maintenance support. However, further evaluation of feasibility, acceptability, and optimal content and dose of such an intervention is needed. Additionally, young adult cancer survivors may be more likely to benefit from eHealth interventions that are not delivered by authority figures.

Introduction

A cancer diagnosis in one's 20s and 30s often results in a host of short and long-term adverse effects, including depression, anxiety, sleep disturbance, fatigue, pain, and loneliness.^1,2 Recent research suggests that behavioral interventions, such as mindfulness training, are feasible for young adult cancer survivors, and may reduce uncertainty, emotional distress, and improve quality of life in this population.^3–5 In fact, the use of mindfulness practices by young adults have steadily increased over the past two decades, suggesting that mindfulness interventions may be well accepted by this population.⁶

A significant challenge in the delivery of mindfulness interventions is assisting participants in using the strategies that have been learned during mindfulness training outside of intervention sessions.⁷ Previous studies have indicated that there may be a link between practice time and psychological outcomes.^8,9 To promote continued mindfulness practice and maintain improvements in quality of life, “booster” treatments of a reduced dosage have been used. Although, at times, participation in such booster treatments are low,¹⁰ a systematic review of mindfulness interventions for patients with bipolar disorder found that the inclusion of a booster session in an intervention was associated with improved maintenance of treatment gains.¹¹

Despite the potential for in-person booster sessions to improve maintenance of outcomes gained during mindfulness training, there are significant logistical barriers to providing such maintenance services, including cost, distance, and personnel availability. In addition, young adult cancer survivors have expressed a preference for interventions that optimize convenience.¹² The delivery of maintenance treatment through eHealth technologies is a low-cost and accessible alternative that may be particularly suited to the needs and preferences of young adult cancer survivors. Ecological momentary interventions (EMIs) through eHealth technologies involve the delivery of intervention content as participants go about their daily lives.¹³ EMIs have been used to provide maintenance support for health behavior change (e.g., alcohol use,¹⁴ smoking cessation,¹⁵ and fruit and vegetable consumption¹⁶), as well as to increase situational awareness and self-monitoring.¹⁷ They have been most consistently efficacious when implemented in combination with other intervention strategies, rather than as a stand-alone intervention.¹³ EMIs are also an ideal intervention to assist participants in transferring what they have learned in an intervention in their daily lives.^7,13 Thus, EMIs may be an ideal strategy to deliver maintenance support to help young adult maintain gains following participation in a mindfulness intervention.

In the present study, we sought to examine attrition rates and preliminary effects of eHealth maintenance support on patient-reported outcomes (PROs) following participation in an in-person mindfulness-based stress reduction (MBSR) intervention. Our hypotheses were:

• 1.Attrition rates in intervention groups would be comparable to the control group.
• 2.Individuals who did not receive weekly mindfulness messages would report declines over time in mindfulness and positive affect, and increases in perceived stress, anxiety, and depression, whereas individuals who received either peer or instructor-framed mindfulness messages would report maintenance of these PROs.

EMI's are typically delivered by study staff or intervention instructors. However, receiving peer support from other survivors has been identified as an important need in young adult cancer survivors.^12,18,19 In addition, peer support has been proposed as a way to increase engagement in online interventions.²⁰ This research has influenced our current study's exploratory question, namely: when delivering EMI messages to maintain study gains with young adult cancer survivors, do young adults with cancer respond differently to the same maintenance messages if they perceive them to be sent from their mindfulness teacher, or instead from another young adult with cancer? Specifically, we wanted to know whether participants were more or less likely to attrite or respond depending on who the message was framed as coming from. In addition, we were curious if there would be differences in maintenance of gains in PROs as a result of message framing. To our knowledge, no studies exist that have compared the impact of this type of message framing of an EMI.

Methods

Ethical procedures

All study procedures and materials were approved by and in accordance with the ethical standards of the Institutional Review Board at the institution where study procedures were conducted. The trial was registered on www.clinicaltrials.gov (NCT02495376).

Recruitment and inclusion/exclusion criteria

Participants were recruited from a large, Midwestern National Cancer Institute (NCI)-designated comprehensive cancer center from June 2014 to June 2016. Participants were recruited through referral from a medical team member, an institutional data repository, and posted brochures and flyers. Upon recruitment, a trained research assistant described the study, answered questions, and conducted informed consent and Health Insurance Portability and Accountability Act authorization to enroll eligible participants.

To participate in the study, participants were required to be between the ages of 18–39 years, have received a cancer diagnosis while within that age range, and speak English well enough for participation in a group of this nature. If patients reported use of mind/body therapies (e.g., mindfulness, yoga, meditation) 3 or more times per week for the past 2 weeks they were not eligible for the study.

Procedures

Three experienced mindfulness teachers drafted an initial list of mindfulness messages that could be emailed or texted to participants. Messages were simple awareness-raising statements and were designed to cue a connection to skills learned in MBSR. These messages were then reviewed and adapted by five young adult cancer survivors who had previously received intensive mindfulness training to ensure their appropriateness for young adults. The messages developed through this process that were ultimately sent to participants are listed in Table 1.

Table 1.

eHealth Messages

Instructor message intro: Here is your weekly mindful message from your mindfulness teacher
Peer message intro: Here is your weekly mindful message from another young adult with cancer
Week	Mindful message
9/25	Remember to STOP from time to time
10/26	In this moment, invite yourself to be just as you are
11/27	Take a bite of something. Close your eyes and use your senses to really notice it
12/28	Remember, a thought is just a thought. Rather than getting caught up in it, let it pass by like a cloud in the sky
13/29	Notice your feet on the ground. Can you be just here in this moment?
14/30	Take in the sounds around you right now
15/31	Recall something pleasant that happened this week. Take yourself back to what you sensed, smelled, saw, and heard. Notice what you feel right now
16/32	Consider sending yourself the following message: May I be free of mental and physical pain, may I be healthy and strong, may I embrace the joy in my life
Closing statement: Please send a “GOT IT” reply to confirm you received this message

STOP, Stop, Take a breath, Observe, Proceed.

As a part of the in-person MBSR parent study, participants were randomized to participate in an 8-week MBSR intervention or a waitlist control group. Findings regarding feasibility, acceptability, and potential intervention effects of in-person MBSR have been published in a separate article.²¹ Most notably, MBSR participants reported significantly greater increases in self-kindness (a subdomain of self-compassion) over time than the waitlist control group. This parent study had a crossover design in which all participants completed assessments at five time points (baseline, 8, 16, 24, and 32 weeks). Group 1 participated in the MBSR course between baseline and 8 weeks, and then received follow-up messages from weeks 8 to 16. Group 2 participated in the MBSR course between 16 and 24 weeks, and then received follow-up messages from weeks 24 to 32 (Fig. 1a).

FIG. 1.

(a) Design of parent study, with information regarding what data were used in the present study. (b) Patient flow diagram. MBSR, mindfulness-based stress reduction.

In the present study, data were combined for the individuals from both Groups 1 and 2. Specifically, data from the intervention group at baseline (pre-MBSR), week 8 (post-MBSR), and week 16 (post-messaging) were combined with data from the waitlist control group at weeks 16 (pre-MBSR), weeks 24 (post-MBSR), and weeks 32 (post-messaging). The only procedural difference between these groups was that the waitlist control group completed two additional assessments before beginning an MBSR group. Initial group randomization was included as a covariate in all analyses.

Following participation in the MBSR group, participants were randomized by a trained research assistant using a computer-generated simple randomization (1:1:1) to 8 weeks of one of the following arms: (1) weekly mindfulness messages sent from the perspective of a mindfulness teacher (N = 24); (2) the same weekly mindfulness messages sent from the perspective of a young adult cancer survivor peer (N = 16); or (3) no messages (N = 22). Sample size was based on the number of participants who completed the MBSR group in the parent study. The messaging groups were unequal solely because that is how randomization schedule played out. Participants randomized to the intervention groups had the option to receive messages through text or email and were informed that they would receive messages once per week for 8 weeks but were not informed of their condition. All participants chose to receive messages through text. Participants were instructed to confirm receipt of their message with a simple reply of “Got it—thanks” and that messaging strategies were not to be used as a method of ongoing communication or for obtaining any additional medical information or advice. Messages were sent by a trained research assistant who was not blinded to participant condition, given that messages were standardized across conditions and did not require any additional contact with study staff.

Study measures

At baseline, participants provided sociodemographic information, clinical characteristics (i.e., cancer type), and completed questionnaires regarding their meditation history and beliefs about meditation. Beliefs about meditation were evaluated using the Expectancy/Credibility Scale,²² which consists of items based on logical thinking and feelings in relation to the therapy being offered. At baseline, weeks 8, and 16 patients completed self-reported outcome measures. Mindfulness was measured using the Mindful Attention and Awareness Scale,²³ a measure of mindful attention that has been validated in a cancer population.²⁴ The Perceived Stress Scale²⁵ was used to examine the degree to which participants appraised events in their life as stressful. Patient-reported outcomes measurement information system (PROMIS) Computer Adaptive Tests were used to measure social isolation,²⁶ emotional support,²⁷ anxiety, depression,²⁸ and positive affect and wellbeing.²⁹

Statistical analyses

Multivariate analyses were used to test for potential group difference between conditions at baseline using Fisher's exact test and one-way analyses of variance. Q–Q plots of outcome variables were visually inspected to evaluate distributions of outcome variables. We tested for intervention effects on each outcome variable from post-MBSR to post-messaging using repeated measures models in a mixed format, which is robust to differences in group size. Each outcome was tested separately as the dependent variable. Message group and time point were entered as fixed effects. Additionally, all models adjusted for meditation history, MBSR randomization, class and retreat attendance during the MBSR portion of the study, and mean baseline (pre-MBSR) score for the outcome being tested.

Planned contrasts were used to further evaluate the impact of intervention group over time on each outcome variable. Specifically, we compared the groups that received a message to the group that did not receive any follow-up messages. Additionally, planned contrasts were used to determine whether there were differences between the group that received messages framed as being from an instructor, and the group that received messages framed as being from a peer. To manage the increased probability of type 1 error due to multiple comparisons for each outcome variable, a Bonferroni correction was applied to the alpha level. As two planned comparisons were performed for each outcome variable, the adjusted alpha level to determine significance for these tests was 0.025.³⁰

Results

Demographic and clinical characteristics

On average, participants were 33.6 years old, and 83.8% had completed at least college. The majority of participants were female (80.6%), Caucasian (88.7%), and non-Hispanic (85.5%). Most participants were either single (43.5%) or married (37.1%) and employed at least part time (87.1%). The most common cancer types were breast cancer (43.5%) and lymphoma (17.7%). Most (81.9%) participants reported that they had never meditated regularly before participation in the MBSR course. On average, participants enrolled in the study 25.82 months (standard deviation = 31.31) after their most recent cancer diagnosis. Additional details about sociodemographic and clinical characteristics by group are presented in Tables 2 and 3; there were no significant differences between groups.

Table 2.

Sociodemographic Characteristics at Baseline

Variable	No message (N = 22), n (%)	Instructor message (N = 24), n (%)	Peer message (N = 16), n (%)	Statistic	p
Age	34.14 ± 3.68	33.79 ± 4.83	32.56 ± 4.95	F(2, 59) = 0.61	0.549
Gender				FET = 0.16	1.000
Male	4 (48.2)	5 (20.8)	3 (18.8)
Female	18 (81.8)	19 (79.2)	13 (81.3)
Race				FET = 3.55	0.524
White	20 (90.9)	20 (83.3)	15 (93.8)
Black/African American	1 (4.5)	4 (16.7)	1 (6.3)
Not provided	1 (4.5)	0 (0.0)	0 (0.0)
Ethnicity				FET = 3.89	0.147
Hispanic	4 (18.2)	1 (4.2)	4 (25.0)
Non-Hispanic	18 (81.8)	23 (95.8)	12 (75.0)
Marital status				FET = 7.20	0.280
Never married	5 (22.7)	14 (58.3)	8 (50.0)
Live-in partner	2 (9.1)	2 (8.2)	1 (6.3)
Married	12 (54.5)	6 (25.0)	5 (31.3)
Divorced	3 (13.6)	2 (8.3)	2 (12.5)
Highest degree earned				FET = 4.17	0.391
High school	1 (4.5)	5 (20.8)	4 (25.0)
College	10 (45.5)	11 (45.8)	6 (37.5)
Graduate	11 (50.0)	8 (33.3)	6 (37.5)
Employment				FET = 2.70	0.708
Employed	20 (90.9)	19 (79.2)	15 (93.8)
Student	1 (4.5)	3 (12.5)	0 (0.0)
Not employed	1 (4.5)	2 (8.4)	1 (6.3)

FET, Fisher's exact test.

Table 3.

Clinical Characteristics at Baseline

Variable	No message (N = 22), n (%)	Instructor message (N = 24), n (%)	Peer message (N = 16), n (%)	Statistic	p
Cancer type				FET = 20.63	0.263
Bone/soft tissue	2 (9.1)	0 (0.0)	0 (0.0)
Brain	0 (0.0)	4 (17.4)	0 (0.0)
Breast	11 (50.0)	7 (30.4)	9 (56.3)
Colorectal	1 (4.5)	0 (0.0)	1 (6.3)
Leukemia	0 (0.0)	3 (13.0)	1 (6.3)
Lymphoma	4 (18.2)	4 (17.4)	3 (18.8)
Melanoma	1 (4.5)	1 (4.3)	1 (6.3)
Stomach/esophageal	1 (4.5)	0 (0.0)	0 (0.0)
Testicular	1 (4.5)	0 (0.0)	0 (0.0)
Thyroid/endocrine	0 (0.0)	2 (8.7)	0 (0.0)
Other	1 (4.5)	2 (8.7)	1 (6.3)
Lymph involvement				FET = 6.25	0.171
Yes	7 (31.8)	13 (54.2)	10 (62.5)
No	10 (45.5)	9 (37.5)	6 (37.5)
Unsure/no response	5 (22.7)	2 (8.3)	0 (0.0)
Distant metastases				FET = 7.95	0.062
Yes	5 (22.7)	1 (4.2)	4 (25.0)
No	16 (72.7)	22 (91.7)	9 (56.3)
Unsure/no response	1 (4.5)	1 (4.2)	3 (18.8)
Recurrence				FET = 5.04	0.216
Yes	6 (27.3)	3 (12.5)	1 (6.3)
No	16 (72.7)	19 (79.2)	15 (93.8)
Unsure/no response	0 (0.0)	2 (8.3)	0 (0.0)
Active treatment				FET = 4.29	0.332
Completed	14 (63.6)	12 (50.0)	6 (37.5)
Ongoing	7 (31.8)	10 (41.7)	10 (62.5)
Unsure/no response	1 (4.5)	2 (8.3)	0 (0.0)
Meditation history				FET = 1.58	0.833
Never	6 (27.3)	9 (39.1)	5 (31.3)
Occasionally	12 (54.5)	11 (47.8)	7 (22.6)
Regularly	4 (36.4)	3 (13.0)	4 (36.4)
Expectancy	0.51 ± 2.98	0.62 ± 2.69	0.13 ± 2.18	F(2, 59) = 0.24	0.790
Credibility	0.58 ± 2.48	0.50 ± 2.04	0.27 ± 2.16	F(2, 58) = 0.13	0.879
MBSR sessions attended	6.59 ± 1.14	6.57 ± 1.47	5.94 ± 1.44	F(2, 59) = 0.93	0.397
Attended MBSR retreat	13 (59.1)	18 (78.3)	14 (87.5)	FET = 3.68	0.151
Message response rate	N/A	7.17 ± 2.06	7.69 ± 0.79	F(1, 38) = 0.94	0.342

MBSR, mindfulness-based stress reduction; N/A, not applicable.

Attrition and response rates

Of the 62 individuals who were randomized to a messaging condition, 60 (96.8%) completed the follow-up assessment. Both participants who left the study between randomization and follow-up reported that they were no longer interested in participating. Although both participants that left the study were in the instructor-framed group, no significant differences in attrition across groups was observed (Fisher's exact test = 2.19, p = 0.332). Similarly, there were no differences between intervention groups in the number of responses to eHealth messages to confirm receipt, t(31.97) = −1.12, p = 0.270.

Intervention effects

Visual inspection of Q–Q plots indicated that all outcome measures were equally distributed. Unadjusted means and standard deviations of PROs are presented in Table 4. There were significant interactions between the messaging group an individual was assigned to and the assessment time point on perceived stress, F(2, 54.73) = 3.92, p = 0.026; anxiety F(2, 56.31) = 5.40, p = 0.007; and depression F(2, 57.39) = 3.67, p = 0.032 after controlling for initial MBSR randomization group, number of MBSR sessions attended, retreat attendance, meditation history, and the baseline (pre-MBSR) level of the outcome variable. In addition, the message group had a main effect on mindful attention and awareness, F(2, 52.73) = 4.07, p = 0.023; anxiety, F(2, 52.04) = 3.20, p = 0.049; and depression, F(2, 52.91) = 3.25, p = 0.047. There was also a statistically significant main effect of time on emotional support, F(1, 58.53) = 4.33, p = 0.042 such that emotional support scores decreased over time regardless of group. No other statistically significant interactions or main effects were observed (Table 5).

Table 4.

Unadjusted Means and Standard Deviations of Outcome Measures

Outcome measure	Pre-MBSR		Post-MBSR		Post-messaging
	Mean	SD	Mean	SD	Mean	SD
No messages
MAAS	3.97	0.98	4.03	0.94	4.28	0.92
PROMIS–Positive Affect	52.84	5.18	53.69	5.25	55.53	6.74
PSS	5.86	2.49	5.65	2.87	4.95	2.71
PROMIS–Anxiety	57.02	8.29	55.53	6.44	53.29	8.02
PROMIS–Depression	50.86	7.70	51.92	6.60	49.79	7.52
PROMIS–Social Isolation	48.12	7.48	50.22	6.80	47.14	10.11
PROMIS–Emotional Support	53.22	8.79	55.69	9.05	54.80	8.13
Instructor-framed messages
MAAS	3.91	0.90	3.50	1.15	3.73	1.10
PROMIS–Positive Affect	52.05	6.50	52.32	7.64	51.61	6.72
PSS	6.12	2.58	5.33	2.58	6.67	2.67
PROMIS–Anxiety	56.25	8.16	56.48	6.92	59.74	8.94
PROMIS–Depression	52.16	9.59	52.31	6.88	56.83	10.17
PROMIS–Social Isolation	51.26	9.04	50.18	7.04	51.56	10.96
PROMIS–Emotional Support	54.51	9.58	56.67	9.14	54.67	7.68
Peer-framed messages
MAAS	3.60	0.81	4.09	0.76	4.19	0.76
PROMIS–Positive Affect	50.14	6.21	52.08	6.97	53.16	4.34
PSS	7.31	2.73	6.00	2.68	5.60	1.99
PROMIS–Anxiety	61.29	6.38	57.50	7.14	54.45	5.27
PROMIS–Depression	56.01	6.71	50.52	9.11	51.33	8.51
PROMIS–Social Isolation	53.47	8.37	47.58	9.98	49.11	7.44
PROMIS–Emotional Support	50.10	6.32	53.92	7.41	51.26	8.40

MAAS, mindful attention and awareness scale; PROMIS, patient-reported outcomes measurement information system; PSS, perceived stress scale; SD, standard deviation.

Table 5.

Within and Between Group Effects from Post-mindfulness-Based Stress Reduction to Post-messaging

Outcome variable source	df	F	p
MAAS
Time	53.09	3.40	0.071
Message group	52.73	4.07	0.023*
Time × message group	53.09	0.03	0.970
PROMIS–positive affect
Time	57.69	0.90	0.347
Message group	54.41	1.46	0.240
Time × message group	57.73	1.71	0.189
PSS
Time	54.71	0.19	0.668
Message group	52.52	0.56	0.572
Time × message group	54.73	3.92	0.026*
PROMIS–anxiety
Time	56.29	0.53	0.470
Message group	52.04	3.20	0.049*
Time × message group	56.31	5.40	0.007*
PROMIS–depression
Time	57.37	1.01	0.320
Message group	52.91	3.25	0.047*
Time × message group	57.39	3.67	0.032*
PROMIS–social isolation
Time	57.06	0.07	0.790
Message group	54.47	1.42	0.251
Time × message group	57.08	1.38	0.259
PROMIS–emotional support
Time	58.53	4.33	0.042*
Message group	54.28	0.35	0.707
Time × message group	58.57	0.37	0.693

Note: All analyses included initial randomization group, number of MBSR sessions attended, retreat attendance, meditation history, and mean baseline (pre-MBSR) value for the outcome variable as covariates.

^*p < 0.05.

Planned contrasts revealed that there were no statistically significant differences between the groups that received eHealth messages and the group that did not (Table 6). However, when comparing the instructor-framed group to the peer-framed group, there was a significant difference in change from post-MBSR to post-messaging anxiety symptoms. Post-hoc univariate comparisons indicated that individuals who received instructor-framed messages reported greater symptoms of anxiety over time, F(1, 56.87) = 5.39, p = 0.024, whereas those who received peer-framed messages did not, F(1, 56.04) = 3.22, p = 0.078 (Fig. 2).

Table 6.

Results of Planned Contrasts of Difference in Change on Outcome Variable from Post-mindfulness-Based Stress Reduction to Post-messaging with Significant Interaction Effects

Outcome variable	Estimate	SE	t	df	p	95% CI		dppc2
						Lower	Upper
No messages vs. messages
PSS	1.15	0.66	1.75	55.15	0.085	−0.17	2.47	0.398
PROMIS–anxiety	2.39	1.83	1.31	56.17	0.196	−1.27	6.06	0.331
PROMIS–depression	4.80	2.20	2.18	57.26	0.033	0.39	9.21	0.530
Instructor-framed vs. peer-framed
PSS	1.56	0.79	1.99	54.25	0.052	−0.01	3.14	0.598
PROMIS–anxiety	6.40	2.23	2.87	56.40	0.006*	1.94	10.87	0.880
PROMIS–depression	3.72	2.69	1.38	57.47	0.172	−1.66	9.10	0.144

^*p < 0.025 (alpha level adjusted for multiple comparisons within each outcome variable using Bonferroni correction).

CI, confidence interval; SE, standard error.

FIG. 2.

Graphs of patient-reported outcomes with significant interaction effects. PSS, perceived stress scale.

Discussion

In this study, we examined attrition rates, response rates, and preliminary effects of eHealth maintenance support following an in-person mindfulness intervention for young adult cancer survivors. In addition, we explored whether peer-framed messages and instructor-framed messages resulted in differential outcomes.

Attrition rates in intervention groups were comparable to the control group—no statistically significant differences in attrition rates were observed between groups. Although this low rate of attrition is encouraging, it is important to note that this was likely heavily influenced by the attrition that occurred during the MBSR intervention as reported in the primary outcomes section of this study.²¹ In addition, there were no differences in response rates between intervention groups. Thus, this study provides initial support for the feasibility of eHealth messages as a means of maintenance support for young adult cancer survivors regardless of the perceived source. However, such rates provide only a limited view of feasibility. Future studies should evaluate additional aspects of feasibility such as time needed to send messages, willingness of participants to receive follow-up messages, and an evaluation of participant's experiences of such messages.

With regard to the preliminary effects of this intervention on PROs, the intervention groups did not demonstrate improved maintenance of PROs compared with the control group. One possible explanation is that the frequency and duration of the intervention (e.g., one weekly message for an 8-week period) was not sufficient. This is consistent with similar findings of eHealth interventions on mindfulness and related outcomes to date. Kraft et al.³¹ used messaging to increase mindfulness practice at home among depressed individuals following discharge from an inpatient psychiatric facility, which did not result in increased mindfulness. Future research should examine the impact of content, frequency, and duration on the efficacy of eHealth follow-up. In addition, regardless of the messaging group, participants reported decreases in emotional support over time. In our study, it is probable that this decline in perceived emotional support was due to the end of the in-person MBSR group. This finding suggests that the eHealth intervention was not sufficient to result in maintenance of perceived emotional support following the MBSR course.

As an exploratory aim, the perceived source of remote follow-up was evaluated to determine if this level of tailoring was important for maintaining improvements in symptoms in this population. Our findings provide preliminary support for this possibility. Specifically, individuals who received messages framed as being from an instructor reported a significant score increase in anxiety symptoms. The estimated marginal mean of the T score on the PROMIS-Anxiety measure for the instructor group increased from 57.55 to 60.91 (3.36 points) between the post-MBSR and post-messaging time points. Established cut scores indicate that this is a shift from mild anxiety (T = 55–60) to moderate anxiety (T = 60–70).³² In contrast, the estimated marginal mean for the peer group decreased from 57.00 to 53.95 (3.05 points). Established cut scores indicate that this is a shift from mild anxiety (T = 55–60) to an anxiety level within normal limits (below T = 55).³² Moreover, the minimally important difference in cancer patients for this scale ranges from 3.0 to 4.5.³³ suggesting that these changes are clinically significant. The clinical significance of our findings are further supported by the effect size (d_ppc2 = 0.880) for this difference in change, which Cohen categorizes as large.³⁴ Based on the results of this study, it is possible that the source of the mindfulness message is quite important; young adult cancer survivors may find it anxiety provoking to receive messages from an instructor, but not a peer. One possible reason for this finding is that participants may have experienced concern about disappointing the mindfulness instructor if they were not as engaged in regular practice as they believed they should be. Future studies should explore cognitive and emotional processes (such a worry about disappointing an authority figure), which may explain the relationship between instructor-framed messages and anxiety symptoms.

This study is not without limitations. Most notably, the sample size was relatively small, with <25 participants per group, as sample size was limited by the number of participants who completed the parent study. Therefore, the present study was likely underpowered for detection of between-group differences. Future studies should seek to recruit larger sample sizes to better characterize the impact of eHealth follow-up. It is also notable that although messages were framed as coming from a peer, they were sent by study staff, and we did not assess whether participants were convinced that messages came from the stated source. It is possible that a stronger effect may be observed if mindfulness messages were more personalized and sent by actual cancer survivor peers. In addition, we did not collect data on patient's existing familial and peer support structures. Thus, this may be a factor that confounds our results. Finally, our sample was quite homogenous with regard to demographic and clinical characteristics. Future studies should also seek to recruit more diverse samples with regard to gender, race, ethnicity, and cancer type.

Conclusions

Overall, the use of EMI following intensive mindfulness training among this population holds promise as a feasible, scalable strategy for the delivery of maintenance interventions. However, in this study, we did not observe any differences between individuals who received messages, and those who did not. Thus, it is not yet clear what the optimal content and dose of an intervention might be to retain participants while also improving maintenance of outcomes. Additionally, in this population, it may be particularly important to consider the source of such an intervention. Young adult cancer survivors may be less likely to benefit from eHealth interventions delivered by authority figures.

Disclaimer

The contents of this article represent original work and have not been published elsewhere in either print or online formats. Some findings from this manuscript have been previously presented as a poster at the 40^th annual meeting and scientific sessions of the society of behavioral medicine.

Author Disclosure Statement

The authors have no conflicts of interest to report, and no competing financial interests exist. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Funding Information

This study was funded by a grant from the American Cancer Society-Illinois Division and a Supportive Care Grant from the Robert H. Lurie Comprehensive Cancer Center (D.V.). K.M.M. was supported by NCI training grants CA193193 and CA122061.