A Randomized Trial of 21 Days of Loving Kindness Meditation for Stress Reduction and Emotional Well-being Within an Online Health Community for Patients,Family,and Friends Experiencing a Cancer Health Journey

Abstract

Objectives:

CaringBridge (CB) is an online health community for people undergoing challenging health journeys. Loving Kindness Meditation (LKM) is a systemized mind–body approach developed to increase loving acceptance and has previously been reported to increase resilience in the face of adversity.

Materials and Methods:

Results of a randomized controlled trial of immediate compared with deferred 21-day LKM intervention in an online community are reported. The deferred group received LKM intervention after a waiting period of 3 weeks. Inclusion criteria were >18 years old, ability to understand English, willingness to participate in a mind–body practice, and use of CB for a cancer journey. Change in perceived stress, self-compassion, social connectedness and assurance, and compassionate love scales from baseline to 21 days was assessed.

Results:

Of the 979 participants included in the study, 649 (66%) provided 3-week follow-up data and 330 (49%) self-reported engaging in the LKM practice 5 or more days/week. Participants in the immediate LKM group reported medium effect size improvement in stress (0.4), self-compassion (0.5), and social connectedness (0.4) compared with the deferred LKM group. Changes in perceived stress and self-compassion were larger in magnitude and increased with more frequent engagement in LKM.

Conclusions:

The immediate LKM group showed improvements in stress, self-compassion, and social connectedness compared with the deferred control group. Differential study retention rates by treatment arm and self-reported engagement in LKM subject the results to selection bias. Future research of similar interventions within online health communities might pay greater attention to promoting intervention adherence and engaging a more diverse economic and racial/ethnic population.

ClinicalTrials.gov (NCT05002842).

Introduction

After a cancer diagnosis, patients and caregivers increasingly turn to online communities as a source of support and help cope with their cancer journey. Research suggests that patients, caregivers, and friends experiencing a cancer journey benefit from the information, personal or professional support, shared experience, and patient advocacy provided by online support groups.^1
–3

CaringBridge (CB) is an online social network specifically aimed at enhancing social connection and support during challenging health journeys. Since its launch in 1997, CB has had 2.4 billion visits and has hosted more than 840,000 participant websites reaching 237 countries and territories worldwide.⁴ The CB platform provides an opportunity for individuals and their caregivers to rally support for a loved one during a health journey. The CB platform offers the opportunity for expressive writing through journal entries, well-being resources, participant health journey stories, and practical scheduling and communication tools to support community, healing, and well-being.

Loving-kindness (Metta in Pali) is a state of unselfish and unconditional kindness to all beings, without exception. Loving Kindness Meditation (LKM) is a kindness-based meditation technique aimed at cultivating loving acceptance toward oneself and all sentient beings.⁵ The LKM follows a systematized approach of directing caring feelings first toward oneself, and then to an expanding circle from oneself to loved ones, acquaintances, strangers, and finally all beings without distinction. Some forms of LKM also include directing loving care toward those with whom one experiences interpersonal difficulties. The meditation practice includes the repetition of short phrases such as “May you be happy, may you be healthy, may you be free from all pain.”

Several systematic reviews and meta-analyses suggest that kindness-based meditation practices such as LKM may be moderately effective in decreasing negative emotions and promoting positive emotions.^6
–8 Galante et al. found that LKM improves self-reported depression, mindfulness, positive affect, compassion, and self-compassion.⁶ Other meta-analyses suggest that LKM improves self-oriented positive emotions⁷ and general psychosocial outcomes.^8,9 In addition, preliminary biological findings and theoretical models suggest that LKM may enhance activation of brain areas around emotional processing and empathy.^10
–12

The potential benefits of LKM extend beyond the general public to those with specific health conditions. The LKM has been shown to decrease pain and psychological distress for those with serious health conditions, such as cancer.^13

–17 For individuals with breast cancer, a preliminary study found that LKM was associated with reduced pain and improved compassion for patients during biopsy and surgery compared with usual care.¹⁸ In addition, there is a potential for LKM interventions to provide benefit for clinical populations suffering from social anxiety, marital conflict, anger, and strains of long-term caregiving.¹⁹

The LKM directed to caregiver populations (e.g., partners, psychologists, nurses, physicians, social workers) indicate feasibility, improved self-compassion and empathic skills of caregivers,^20,21 and acceptance of online delivery of LKM.^22,23 A previous internet-delivered intervention of brief daily gratitude practice specific to the CB population confirmed the feasibility of recruitment of a large number of participants (patients, caregivers, family, and friends) experiencing a health journey and web-based delivery of a brief mind–body intervention.²⁴

The purpose of this article is to report the findings of a prospective randomized controlled trial comparing immediate LKM with a deferred intervention. Participants were provided a pre-recorded, brief LKM via e-mail and were instructed to practice the LKM daily for 3 weeks. The impact of immediate versus delayed LKM on perceived stress, self-compassion, social support, social connectedness, and compassionate love among a population experiencing a cancer journey was assessed.

Materials and Methods

Design and participants

This study randomized CB users that were part of a cancer journey to either an immediate or a deferred 21-day LKM. The Institutional Review Board at the University of Minnesota (UMN) approved this research (1703S09842). The study was retroactively registered with ClinicalTrials.gov.

Participants were recruited from May 2018 to September 2018 via www.caringbridge.org. The CB users were made aware of the potential benefits of short, mind–body interventions such as LKM and were offered the opportunity to participate. The CB users were given the option to participate via a link on the CB site, and a Facebook ad on the CB Facebook site featuring a video briefly explaining the purpose of the study. Further, approximately 5000 e-mails were sent by CB to highly engaged CB subscribers undergoing a cancer journey (i.e., individuals associated with a cancer journey—their own or as a friend/caregiver) to solicit interest in study participation. Inclusion criteria were: adults 18 years of age or older, ability to read and write in English, willingness to participate in a mind–body practice, active user of CaringBridge.org with a valid e-mail address, and visitors, caregivers, or patient-authors to a cancer journey page.

Those who responded to recruitment solicitation were directed to a study website hosted by the Earl E. Bakken Center for Spirituality and Healing at the UMN. Participants consented electronically and were screened online, via the study website. Individuals who met all inclusion criteria, completed the initial study survey, and consented to participate were enrolled into the study and randomized (virtually) to one of the study interventions with an automated, concealed process by using the REDCap Randomization module.²⁵

Study intervention

Enrolled participants, eligible individuals who consented and completed the baseline survey, were randomly placed into an immediate intervention group or an intervention group that began 21 days later (deferred) by using a 1:1 concealed, allocation ratio with randomly permuted blocks of size 2 and 4 generated by the study statistician using Stata 15.1. Participation in this study was 21 days for those in the immediate LKM group and 42 days for those in the group that starts LKM after a 21-day waiting period (21 days of regular interaction with CB and 21 days with LKM). There was no cost to participate. All participants were e-mailed a link to the study website, which included an online survey at the beginning of the study. On completion of the survey, eligible participants were enrolled into the study and randomized to an unblinded intervention group.

Blinding to group assignment was not possible given the nature of the intervention and self-assessment of outcomes. Participants in the immediate intervention group received simple instructions and a link to a website with a 12-min audio guided recording of the LKM practice (Appendix A1) via e-mail.

Instructions for the meditation included closing one's eyes or allowing one's gaze to relax, attention to feelings and body sensations, and directed thoughts with simple phrases to silently repeat (e.g., May I be happy. May I be well. May I live with ease and joy.) Participants in the deferred intervention group received an e-mail message thanking them for their participation along with an explanation on the importance of participating in the control arm, and a promise of future contact with instructions for LKM in 21 days. Participants in both groups received automated prompts via e-mail, once per week, during active LKM practice weeks to thank them for participating and reminding them to do their practice. The prompts included the instructions and a link to the study website; participants could refer to these anytime during the study.

Outcome measures

Study participants in both groups completed self-report questionnaires at baseline and 3 weeks. Participants in the deferred group also completed self-report outcomes after they completed the LKM intervention (at 6 weeks). Surveys were administered to individual participant e-mail addresses by using REDCap.²⁵

Participants received up to three automated requests to complete the follow-up survey through REDCap. The survey included demographic questions and five outcome measures: perceived stress, self-compassion, social support, social connectedness, and compassionate love. Perceived stress was the primary outcome, and estimates for sample size calculations from a similar study evaluating gratitude²⁴ on change in perceived stress were used to ensure an adequately powered study. Safety was not explicitly evaluated by using a study instrument. Individuals withdrew from the study via e-mail, and a request for reason for withdrawal was made once in reply via e-mail.

Demographics included age, gender, race, ethnicity, household income, religious/spiritual practice, marital status, education, community, health journey in addition to cancer (e.g., hospice care, surgery), and previous experience with mind–body practices. Participants were asked to define their user experience with CB as either a Patient-Author (a patient facing a cancer journey, who created a site, or provided most updates about their health journey); Caregiver-Author (a caregiver to another person facing a cancer journey, who created the site, and/or provides most updates about their loved one's journey); Visitor (a person who follows another's cancer journey and reads Patient-Author or Caregiver-Author journal entries, and/or posts photos or greetings, thoughts, etc.).

Perceived stress scale (PSS) is a broadly used 10-item scale for measuring perception of stress with acceptable reliability and validity (range 0–40).²⁶ Higher PSS scores are associated with greater vulnerability to stressful life-event-elicited depressive symptoms.²⁷ The PSS scores were obtained by summing across all scale items after reverse coding any positively phrased scales. All 10 items are assessed by asking how often a respondent felt a certain way on a 5-point Likert measure (0 = Never; 4 = Very Often).

Self-compassion scale (SCS) is a reliability-tested, 12-item shortened version of the original longer 26-item SCS (range 12–60).^28,29 The 12-item SCS contains the same higher-order factor structure as the 26-item scale with a General higher-order self-compassion factor and six second-order factors of Self-Kindness, Self-Judgment, Common Humanity, Isolation, Mindfulness, and Over-Identification. Higher SCS scores are associated with psychological well-being and suggest that self-compassion might be an important protective factor, fostering emotional resilience.³⁰ An overall SCS was derived by summing across all scale items after reverse coding any negatively phrased scales. All 12 items are assessed by asking how often a respondent felt a certain way on a 5-point Likert measure (1 = Almost never; 5 = Almost always).

Social connectedness and social support measures were evaluated with the social connectedness and social assurance scales, which have been shown to be highly reliable and valid measures of belongingness (both with range 8–48).³¹ Social connectedness and social support are each assessed with 8 items on a 6-point Likert scale (1 = strongly agree through 6 = strongly disagree) by summing across responses to items. Social connectedness was reverse coded such that high scores indicate a higher social connection.

Compassionate love scale (CLS) is a 21-item measure of the stranger–humanity version of the original CLS designed to measure compassionate love for humanity (range 38–147).³² Higher CLS scores indicate greater compassionate love for humanity. The CLS scores were obtained by summing across all 21-scale items. Each item was assessed on a 7-point Likert-type scale from 1 (not at all true of me) to 7 (very true of me). During scale development, testing of psychometrics among undergraduate students resulted in an overall Cronbach's alpha = 0.95.³²

Statistical analysis

Sample size calculations from a similar study evaluating gratitude²⁴ on change in perceived stress suggested estimated loss of follow-up of 47% and standard deviation = 5.7 points. Using 80% power, a 5% two-sided type I error mean difference in perceived stress between groups of 1.5 points can be detected with an enrollment of n = 970 participants. Descriptive analyses were performed to describe the sample and evaluate modeling assumptions. Linear regression was used to evaluate differences between intervention groups on changes from baseline to 21 days in stress, self-compassion, social connectedness, social assurance, and compassionate love. The analysis was intention-to-treat and required a completed instrument, which varied by outcome and controlled for demographic variables.

Cohen's d effect sizes were calculated by using unstandardized regression coefficients of the intervention group as the numerator and estimates the within-group pooled standard deviation as the denominator.³³ We used adjusted linear regression to evaluate the potential for a dose effect of self-reported engagement with LKM practice (days/week) on change in perceived stress, self-compassion, social connectedness, social assurance, and compassionate love for all participant completers. For this analysis only, completers are defined as having baseline and 21-day measures for the immediate group, and 21- and 42-day measures for the deferred group. Dose effect models were adjusted for demographics and randomization group (immediate, deferred) to control for time period effects.

Loess curve plots were used to investigate potential non-linear relationships between days per week of self-reported engagement in LKM practice and the outcomes. When indicated, potential non-linear relationships were tested by including higher-order terms (e.g., quadratic) for days/week of engagement within the adjusted linear regression model. All regression models use Huber-White robust standard errors.

Sensitivity analysis included: A comparison of baseline demographics and outcomes between individuals completing at least one 21-day follow-up outcome and those who did not was performed by using chi-square test of independence and student's t-tests. All analyses were repeated by using inverse probability weighting (IPW). The IPW reflects what the dataset would be like if all participants provided complete data and data are missing at random. All analyses were performed in Stata 15.SE (College Station, TX).

Results

Participant flow

A total of 1978 were screened for the study, and 1921 participants electronically consented to participate in the study over 4 months of recruitment (Fig. 1). A total of 239 individuals did not meet eligibility criteria, and n = 703 did not complete the initial study survey. A total of 979 individuals were enrolled in the study, with 489 randomly allocated to the immediate intervention group and 490 to the deferred intervention group. A total of 10 participants withdrew from the study after randomization (n = 6 immediate, n = 4 deferred). Reasons for withdrawal included major life events and an aversion to the tenor of the recorded voice for the LKM recording. Follow-up data at 21 days on at least one of the completed outcomes measures were collected for 649 participants (66.3%) (immediate = 276 [56.4%], deferred = 373 [76.1%]).

FIG. 1.

Participant flow.

For the deferred group only, 232 participants completed outcomes at day 42 (47.3% of those randomized and 65.2% of those completing outcomes at day 21). Among the participants who completed surveys after their LKM intervention period (n = 254), 49.4% self-reported practicing LKM 5 or more days per week (50.4% immediate; 48.4% deferred). Approximately 17% of participants reported listening to the audio recording on a daily basis as was requested.

Participant demographics

Most participants reported user experience as visitors (62.7%) to a site. The majority of study participants were female (94.1%), white (91.0%), and Christian (51.9%) with higher socioeconomic status (42.5% with incomes $100,000 or above, 75.3% with bachelor's degree or above). An additional health journey in addition to cancer was reported for 33.5% of participants, and more than half of the participants (58.3%) were older than 55 years. More than half (55.4%) of the participants had participated in the past in a mind–body practice. Geographically, the sample included participants from rural (23.2%), urban (27.7%), and suburban (49.2%) settings (Table 1).

Table 1.

Demographic Characteristics by Intervention Group

Characteristics	Immediate N = 489	Deferred N = 490	Characteristics	Immediate N = 489	Deferred N = 490
User experience			Education
Patient-author	98 (20.3)	100 (20.6)	High school graduate or less	72 (14.8)	67 (13.7)
Caregiver-author	85 (17.6)	79 (16.3)	Associate/technical/trade degree	54 (11.1)	48 (9.4)
Visitor	301 (62.2)	307 (63.2)	Bachelor's degree	150 (30.7)	137 (28.1)
Health journey			Graduate degree or higher	212 (43.4)	236 (48.4)
Cancer only	321 (65.6)	330 (67.4)	Household income
Cancer with additional health journey	168 (34.4)	160 (32.7)	Less than $25,000	32 (7.7)	29 (7.04)
Age (years)			$25,000–$34,999	22 (5.3)	23 (5.6)
18–24	3 (0.6)	5 (1.0)	$35,000–$49,999	37 (8.9)	28 (6.8)
25–34	17 (3.5)	19 (3.9)	$50,000–$74,999	78 (18.8)	78 (18.9)
35–44	61 (12.6)	51 (10.4)	$75,000–$99,999	68 (16.4)	80 (19.4)
45–54	124 (25.6)	126 (25.38)	$100,000–$149,999	97 (23.4)	87 (21.1)
55–64	136 (28.0)	173 (35.4)	$150,000–$199,999	42 (10.1)	43 (10.4)
65–74	124 (25.6)	98 (20.0)	$200,000 or more	38 (9.2)	44 (10.7)
75 or older	20 (4.1)	17 (3.5)	Community
Gender			Urban	130 (26.9)	139 (28.5)
Male	29 (6.0)	29 (5.9)	Suburban	235 (48.6)	243 (49.8)
Female	458 (94.0)	461 (94.1)	Rural	119 (24.6)	106 (21.7)
Race/ethnicity			Spiritual worldview
White	445 (91.0)	446 (91.0)	Christian	225 (52.2)	253 (51.6)
Black	5 (1.0)	9 (1.8)	More than one worldview	96 (19.6)	90 (18.4)
Hispanic	9 (1.8)	15 (3.1)	Spiritual but not religious	60 (12.3)	63 (12.9)
Asian	7 (1.4)	5 (1.0)	Other^a	78 (16.0)	84 (17.1)
Other^b	23 (4.7)	15 (3.1)	Previous mind–body experience
Marital status			No	228 (47.4)	203 (41.7)
Single, never married	47 (9.7)	46 (9.5)	Yes	253 (52.6)	283 (58.2)
Married/partnered	350 (72.2)	335 (68.9)
Widowed	32 (6.6)	38 (7.8)
Divorced/separated	56 (11.6)	67 (13.8)

Agnostic, Animist, Atheist, Baha'i, Buddhist, Deist, Hindu, Humanist, Jewish, Muslim, Pagan, Pantheist, Polytheist, Secular, Sikh, Taoist, Unitarian/Universalist, Wiccan, Other, or selected “Prefer not to answer.”

American, Pacific Islander, indicated multiple races, or selected “Prefer not to answer.”

Outcomes

After 21 days, the group receiving immediate LKM reported decreases in perceived stress (−3.3 on 0–40 scale) and increased self-compassion (3.6 on 12–60 scale), social connectedness (2.3 on 8–48 scale), and social assurance (1.3 on 8–48 scale). The deferred LKM group reported decreased perceived stress (−0.6 on 0–40 scale) and compassionate love (−3.6 on 38–147 scale) as well as increased social assurance (0.6 on 8–48 scale). Relative to the deferred group, the immediate LKM group had significantly larger 21-day changes in perceived stress (mean difference of −2.7; 95% CI −3.8 to −1.6), self-compassion (mean difference 3.5; 95% CI 2.2–4.7), and social connectedness (mean difference 2.4; 95% CI 1.1–3.7). Standardized effect sizes for all significant group differences were medium in magnitude (0.4 [perceived stress and social connectedness], 0.5 [self-compassion]; p < 0.001) (Table 2).

Table 2.

Baseline and Adjusted Changes from Baseline to 21-Day Follow-Up in Perceived Stress, Self-Compassion, Social Connectedness, Social Assurance, and Compassionate Love Comparison by Intervention Group

Outcomes	Immediate				Deferred				Difference (95% CI)	Effect size^a (95% CI)	p
	Baseline		21-Day follow-up		Baseline		21-Day follow-up
	N	Average (sd)	N	Post minus pre (95% CI)	N	Average (sd)	N	Post minus pre (95% CI)
Perceived stress (0–40)	478	18.2 (7.5)	258	−3.3 (−4.2 to −2.5)	485	18.1 (7.3)	356	−0.6 (−1.2 to 0.0)	−2.7 (−3.8 to −1.6)	0.4 (0.3 to 0.6)	<0.001
Self-compassion (12–60)	476	40.4 (9.7)	255	3.6 (2.7 to 4.5)	471	40.4 (9.3)	336	0.1 (−0.6 to 0.9)	3.5 (2.2 to 4.7)	0.5 (0.4 to 0.7)	<0.001
Social Connectedness (8–48)	485	39.0 (9.6)	256	2.3 (1.3 to 3.2)	486	38.2 (9.9)	343	−0.1 (−0.8 to 0.6)	2.4 (1.1 to 3.7)	0.4 (0.2 to 0.5)	<0.001
Social assurance (8–48)	483	36.9 (7.9)	253	1.3 (0.5 to 2.1)	479	37.1 (7.3)	339	0.6 (0.0 to1.3)	0.7 (−0.4 to 1.8)	0.1 (−0.1 to 0.3)	0.2
Compassionate love (38–147)	471	107.8 (21.1)	238	−1.9 (−3.9 to 0.1)	461	106.2 (20.4)	326	−3.6 (−5.1 to −2.0)	1.7 (−1.0 to 4.4)	0.1 (−0. to 0.3)	0.2

Adjusted for user experience, health journey, age category, gender, race/ethnicity, marital status, education, household income, community, spiritual worldview, and previous mind–body experience.^aCohen's d effect sizes were calculated by using unstandardized regression coefficients of the intervention group as the numerator and estimates the within-group pooled standard deviation as the denominator.

Average changes in perceived stress showed a quadratic trend for a dose effect with a peak reduction at 7 days (crude change = −4.8 points; p < 0.001) when adjusting for intervention group and baseline demographics. Similarly, there were increases in self-compassion peaks at 6 days (adjusted average change = 5.5; p = 0.003). There did not appear to be a pattern for social connectedness, social assurance changes, and compassionate love with regard to the number of days of self-reported engagement in LKM practice (Table 3).

Table 3.

Among Completers Among Both Intervention Groups, Effect of Self-Reported Engagement (Days/Week) with Loving Kindness Meditation Practice (Days/Week) on Adjusted Average 21-Day Change of Perceived Stress, Self-Compassion, Social Connectedness, Social Assurance, and Compassionate Love

Days	n (%)	Perceived stress		Self-compassion		Social connectedness		Social assurance		Compassionate love
		Ave pre^*	Ave change	Ave pre	Ave change	Ave pre	Ave change	Ave pre	Ave change	Ave pre	Ave change
		95% (CI)	95% (CI)	95% (CI)	95% (CI)	95% (CI)	95% (CI)	95% (CI)	95% (CI)	95% (CI)	95% (CI)
0	44 (8.6)	19.5 (17.0 to 22.1)	0.0^b (−2.2 to 2.2)	39.9 (36.9 to 42.9)	1.1^a (−0.9 to 3.1)	40.5 (38.0 to 43.0)	−0.6 (−2.9 to 1.7)	37.4 (34.7 to 40.0)	0.6^ab (−1.4 to 2.5)	114.4 (108.6 to 120.1)	−3.6 (−7.1 to 0.0)
1	58 (11.3)	17.7 (15.8 to 19.6)	-1.8 ^ab (−3.3 to −0.2)	40.0 (37.2 to 42.8)	1.8 ^ab (0.2 to 3.4)	38.4 (35.7 to 41.1)	0.9 (−0.9 to 2.7)	36.4 (34.1 to 38.8)	3.7 ^b (1.5 to 6.0)	109.1 (103.3 to 114.8)	−4.4 (−9.7 to 1.0)
2	36 (7.0)	15.2 (13.1 to 17.2)	0.4^b (−1.9 to 2.6)	41.6 (38.0 to 45.2)	1.3^ab (−1.3 to 4.0)	35.9 (31.8 to 40.0)	2.5 (−0.5 to 5.4)	37.2 (34.8 to 39.7)	0.0 (−2.0 to 1.9)	105.0 (97.3 to 112.7)	−3.7 (−9.0 to 1.6)
3	67 (13.0)	15.6 (13.8 to 17.5)	−1.5^ab (−3.1 to 0.2)	42.8 (40.3 to 45.2)	2.1 ^ab (0.6 to 3.6)	41.2 (39.0 to 43.4)	0.3 (−1.5 to 2.1)	37.7 (36.1 to 39.4)	1.4^ab (0.0 to 2.8)	104.8 (99.2 to 110.4)	0.9 (−2.2 to 4.1)
4	55 (10.7)	16.7 (14.9 to 18.6)	−3.9 ^ab (−5.8 to −2.0)	42.3 (39.9 to 44.7)	3.3 ^ab (1.2 to 5.3)	39.1 (36.4 to 41.8)	0.8 (−1.2 to 2.8)	38.0 (36.1 to 39.9)	1.6^ab (−0.1 to 3.2)	108.7 (102.6 to 114.9)	−2.0 (−5.9 to 2.0)
5	76 (14.8)	15.9 (14.2 to 17.7)	−2.9 ^ab (−4.4 to −1.3)	42.2 (40.0 to 44.4)	3.9 ^ab (2.3 to 5.5)	39.3 (37.3 to 41.2)	2.1 (0.5 to 3.7)	37.0 (35.4 to 38.7)	1.6 ^ab (0.4 to 2.7)	106.8 (101.6 to 112.1)	−0.2 (−3.4 to 3.0)
6	90 (17.5)	17.1 (15.6 to 18.5)	−4.5 ^a (−5.6 to −3.3)	39.7 (37.5 to 41.8)	5.5 ^b (3.9 to 7.1)	39.0 (37.2 to 40.9)	2.7 (1.3 to 4.2)	37.9 (36.3 to 39.5)	1.3^ab (−0.1 to 2.7)	103.6 (99.1 to 108.2)	4.0 (0.7 to 7.4)
7	88 (17.1)	17.4 (15.8 to 19.3)	−4.8 ^a (−6.2 to −3.4)	41.3 (38.9 to 43.7)	4.8 ^ab (3.3 to 6.4)	38.7 (36.4 to 40.9)	2.4 (0.7 to 4.2)	38.7 (37.2 to 40.1)	−0.8^a (−2.2 to 0.5)	105.0 (100.1 to 110.0)	0.1 (−3.2 to 3.5)
p-Value for the quadratic trend^†			<0.001		0.003		0.2		0.03		0.04

Completers are defined as having baseline and 21-day measures for immediate group, and 21- and 42-day measures for the deferred group. Adjusted average pre and changes in outcomes by self-reported engagement estimated from generalized linear model using robust standard errors with significant difference (α = 0.05) in bold for change. Adjusted for baseline intervention group, age group, gender, user experience, race/ethnicity, health journey, education, household income, spiritual worldview, marital status, community, and previous mind–body experience. For post hoc tests within columns, margins sharing a letter in the group label are not significantly different at the 5% level by using Dunn-Sidak multiplicity correction.

Pre is baseline for the immediate group and 21-day measure for the deferred group.

†

p-Values for change measures are for quadratic trend. There was no statistically detectable linear or quadratic trend for pre measures.

Sensitivity analysis of baseline demographics comparison between individuals completing at least one 21-day follow-up outcome and those who did not indicated that participants with previous mind–body experience were more likely to complete follow-up (p < 0.001). In addition, baseline perceived stress score of participants for non-completers is 19.4 and for completers is 17.5 (p < 0.001). Self-compassion scores are 39.3, non-completers and 40.8, completers (p = 0.02), and social assurance scores are 35.9, non-completers and 37.6, completers (p = 0.001).

There is no statistical evidence of differences in baseline social connectedness and compassionate love between completers and non-completers. Compassionate love scores differed by completer and intervention group combination (p = 0.04) but not after adjustment for multiple comparisons. Sensitivity analyses suggests there might be potential for collider stratification bias because completion rates are differential by intervention assignment. No other outcome differences by intervention group resulted in statistical differences (Table 4).

Table 4.

Baseline Perceived Stress, Self-Compassion, Social Connectedness, Social Assurance, and Compassionate Love Comparison by Intervention Group, by 21-Day Follow-Up, and by Intervention Group 21-Day Follow-Up Combination

Baseline measures	Participants with baseline survey only (non-completers)				Participants with baseline and 21-day follow-up survey (completers)				p^a	p^b
	Immediate		Deferred		Immediate		Deferred
	N	Average (sd)	N	Average (sd)	N	Average (sd)	N	Average (sd)
Perceived stress (0–40)	204	19.3 (7.6)	115	19.6 (7.7)	274	17.4 (7.4)	370	17.6 (7.4)	<0.001	0.9
Self-compassion (12–60)	203	39.0 (9.7)	109	40.0 (10.4)	273	41.4 (9.6)	362	40.4 (8.9)	0.02	0.2
Social connectedness (8–48)	211	38.5 (9.7)	115	36.4 (11.2)	274	39.4 (9.5)	371	38.8 (9.4)	0.06	0.3
Social assurance (8–48)	209	35.6 (8.5)	113	36.4 (8.5)	274	37.9 (7.4)	366	37.3 (6.9)	0.001	0.2
Compassionate love (38–147)	204	106.3 (22.3)	108	109.9 (21.4)	267	108.9 (20.2)	353	106.2 (20.0)	0.9	0.04

p-Value for t-test of means by follow-up (e.g., baseline survey only compared with baseline and 21-day follow-up survey).

p-Value from generalized linear model with robust standard errors for means by intervention by follow-up (e.g., comparison of fours groups: baseline survey only immediate, baseline survey only deferred, baseline and 21-day follow-up survey immediate, baseline and 21-day follow-up survey deferred).

All analyses were repeated by using IPW and did not result in consequential differences in values compared with non-IPW analyses (results not shown).

Discussion

Online health communities provide opportunities for peer-to-peer interaction to share information, personal experience and support, advocacy, and practical scheduling and communication tools to promote healing and well-being.^34
–36 In addition, online health communities provide a means for health research professionals to reach potential study participants, conduct scientific research, and provide evidence-based resources for self-care to consumers.^37
–39 However, more rigorous research is needed to fully understand the potential for health professionals to positively support individuals utilizing online health communities.⁴⁰

To our knowledge, this is one of the first randomized studies to examine the use and potential impact of a brief LKM practice in a fully online health network of those undergoing a cancer journey. Our findings provide insight on who may engage in online health-related self-care research and LKM practice specifically. These findings can be used to inform future design, and implementation of brief self-care interventions to support health in online settings.

Medium favorable effect size changes (0.4–0.5) were observed for stress, self-compassion, and social assurance between the immediate and deferred intervention groups. Decreased stress and increased self-compassion appear to be responsive to self-reported engagement in LKM practice for the first 4 days and appear to plateau afterward without statistically detectable increases or decreases beyond. The observed medium effect sizes, similar to what has been observed in much of the LKM research to date,^6
–8 may be a consequence of our study population and subsequent compliance to engagement in practice and follow-up survey assessment.

Observed baseline levels of stress for participant completers (17.5) are only slightly higher than what is considered normal for the population as a whole. In 2009, 15.5 was the average for men in this age range and 16.1 was the average for women.⁴¹ Moreover, participants with higher levels of stress are less likely to be completers, resulting in the selection bias of study participation toward individuals with lower stress. Future studies might evaluate the acceptability of standard LKM practice for individuals with high stress.

The online platform, a tool to reach participants, conducts virtual randomization and delivers an LKM intervention, which has demonstrated important potential for research in online health communities. Although enrollment and virtual randomization was successful (nearly 1000 participants in more than 4 months, aged 18–75 years and older from rural, suburban, and urban communities with a variety of spiritual belief systems), additional attention is needed to reach a more diverse participant population. This may include revisions to recruitment strategies, and study engagement materials including the LKM script for sensitivity to social and cultural differences, as well as changes in study design to include outcome measures appropriately validated in diverse populations are necessary.

Importantly, a little less than one-fifth of participants reported listening to the LKM on a daily basis as instructed, which may have led to an under-estimation of the intervention effect given the previous research of Fredrickson et al., which supports a dose response between longer average lengths of LKM practice and higher average levels of daily positive emotions.⁴² Given the apparent dose effect on perceived stress and self-compassion, more effort should be devoted to systematically monitoring and encouraging adherence to LKM to facilitate intervention fidelity. Finally, increased efforts are required to improve study adherence and reduce loss to follow-up, including phone calls/texts, personalized e-mail reminders, and compensation for study participation.

Limitations

Recruitment efforts were electronic (e.g., banners, e-mails) and broad (e.g., broadcast via website banner) within the CB network. As such, those who elected to participate are similar temporally in demographics to the CB network and those who chose to participate are likely different than those who did not participate in the study. The study population was homogenously female, white, Christian, and of high socioeconomic status and is not representative of the majority of people undergoing cancer journeys in the population.

The loss to follow-up (33.7%) was high despite participants receiving multiple automated reminders to complete the post-intervention surveys and differentially higher by 20% in the deferred intervention group compared with the immediate intervention group, resulting in the potential for selection bias. Baseline stress, self-compassion, social connectedness, and social assurance were not extreme and were more favorable for those who completed both baseline and the 21-day survey (completers) compared with those with only baseline survey results (non-completers), suggesting participants with follow-up were different than non-completers. For future studies of online interventions such as these, additional efforts should be considered to enroll more economic and racially diverse participants and efforts must be made to enhance compliance.

Conclusions

This study demonstrated the potential of online health communities for recruitment and implementation of rigorous scientific research of brief mind–body interventions. Participants who engaged in a 12-min daily audio-recorded LKM intervention reported statistically significant improvements in stress and self-compassion compared with a deferred control group, with larger changes occurring with more frequent engagement in LKM practice. Due to the limited diversity of the study population and significant lost to follow-up rate, caution is needed with interpretation of study results. Future research requires identifying ways to engage a more diverse economic and racial/ethnic population in research within online health communities.

Footnotes

Acknowledgments

The Implementation of Mind-Body Practices in an Online Community of Caring (CaringBridge) is a team effort and could not have been accomplished without the dedicated staff at CaringBridge. The authors thank their study participants who generously shared their personal experience and health journeys with the University of Minnesota Research Staff.

Disclaimer

The content is solely the responsibility of the authors and does not necessarily represent the official views of CaringBridge.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

Research is supported by CaringBridge (PI: MaryJo Kreitzer).

References

Klemm

, Bunnell

, Cullen

, et al. Online cancer support groups: A review of the research literature. Comput Inform Nurs, 2003; 21:136–142.

Beaudoin

, Tao

. Benefiting from social capital in online support groups: An empirical study of cancer patients. Cyberpsychol Behav. 2007; 10:587–590.

Coulson

, Greenwood

. Families affected by childhood cancer: An analysis of the provision of social support within online support groups. Child Care Health Dev, 2012; 38:870–877.

CaringBridge. 2020. Online document at: https://www.caringbridge.org/about-us/ Accessed May 30, 2020 .

Salzberg

Lovingkindness: The Revolutionary Art of Happiness: Shambhala Publications Inc. Boulder, Colorado, 1995.

Galante

, Galante

, Bekkers

, Gallacher

. Effect of kindness-based meditation on health and well-being: A systematic review and meta-analysis. J Consult Clin Psychol, 2014; 82:1101–1114.

Zeng

, Chiu

, Wang

, et al. The effect of loving-kindness meditation on positive emotions: A meta-analytic review. Front Psychol, 2015; 6:1693.

Shonin

, Gordon

, Compare

, et al. Buddhist-derived loving-kindness and compassion meditation for the treatment of psychopathology: A systematic review. Mindfulness, 2015; 6:1161–1180.

Luberto

, Shinday

, Song

, et al. A systematic review and meta-analysis of the effects of meditation on empathy, compassion, and prosocial behaviors. Mindfulness (N Y), 2018; 9:708–724.

10.

Mascaro

, Darcher

, Negi

, Raison

. The neural mediators of kindness-based meditation: A theoretical model. Front Psychol, 2015; 6:109.

11.

Lutz

, Brefczynski-Lewis

, Johnstone

, Davidson

. Regulation of the neural circuitry of emotion by compassion meditation: Effects of meditative expertise. PLoS One, 2008; 3:e1897.

12.

Germer

, Siegel

. Wisdom and Compassion in Psychotherapy: Deepening Mindfulness in Clinical Practice. New York, NY: Guilford Press, 2012.

13.

Tonelli

, Wachholtz

. Meditation-based treatment yielding immediate relief for meditation-naive migraineurs. Pain Manag Nurs, 2014; 15:36–40.

14.

Carson

, Keefe

, Lynch

, et al. Loving-kindness meditation for chronic low back pain: Results from a pilot trial. J Holist Nurs, 2005; 23:287–304.

15.

Kearney

, Malte

, McManus

, et al. Loving-kindness meditation for posttraumatic stress disorder: A pilot study. J Trauma Stress, 2013; 26:426–434.

16.

Johnson

, Penn

, Fredrickson

, et al. A pilot study of loving-kindness meditation for the negative symptoms of schizophrenia. Schizophr Res, 2011; 129:137–140.

17.

Lang

, Malaktaris

, Casmar

, et al. Compassion meditation for posttraumatic stress disorder in veterans: A randomized proof of concept study. J Trauma Stress, 2019; 32:299–309.

18.

Wren

, Shelby

, Soo

, et al. Preliminary efficacy of a lovingkindness meditation intervention for patients undergoing biopsy and breast cancer surgery: A randomized controlled pilot study. Support Care Cancer, 2019; 27:3583–3592.

19.

Hofmann

, Grossman

, Hinton

. Loving-kindness and compassion meditation: Potential for psychological interventions. Clin Psychol Rev, 2011; 31:1126–1132.

20.

Bibeau

, Dionne

, Leblanc

. Can compassion meditation contribute to the development of psychotherapists' empathy? A review. Mindfulness, 2016; 7:255–263.

21.

Boellinghaus

, Jones

, Hutton

. The role of mindfulness and loving-kindness meditation in cultivating self-compassion and other-focused concern in health care professionals. Mindfulness, 2014; 5:129–138.

22.

Rao

, Kemper

. Online training in specific meditation practices improves gratitude, well-being, self-compassion, and confidence in providing compassionate care among health professionals. J Evid Based Complementary Altern Med, 2017; 22:237–235.

23.

Kohle

, Drossaert

CHC

, Jaran

, et al. User-experiences with a web-based self-help intervention for partners of cancer patients based on acceptance and commitment therapy and self-compassion: A qualitative study. BMC Public Health, 2017; 17:225.

24.

Kreitzer

, Telke

, Hanson

, et al. Outcomes of a gratitude practice in an online community of caring. J Altern Complement Med, 2019; 25:385–391.

25.

Harris

, Taylor

, Thielke

, et al. Research electronic data capture (REDCap)—A metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform, 2009; 42:377–381.

26.

Lee

EH.

Review of the psychometric evidence of the perceived stress scale. Asian Nurs Res (Korean Soc Nurs Sci), 2012; 6:121–127.

27.

Cohen

, Kamarck

, Mermelstein

. A global measure of perceived stress. J Health Soc Behav, 1983; 24:385–396.

28.

Neff

KD.

The development and validation of a scale to measure self-compassion. Self Identity, 2003; 2:223–250.

29.

Raes

, Pommier

, Neff

, Van Gucht

. Construction and factorial validation of a short form of the Self-Compassion Scale. Clin Psychol Psychother, 2011; 18:250–255.

30.

Leary

, Hoyle

, ebrary

. Handbook of Individual Differences in Social Behavior. New York, NY: Guilford Press, 2009.

31.

Lee

, Robbins

. Measuring belongingness: The social connectedness and the social assurance scales. J Couns Psychol, 1995; 42:232–241.

32.

Sprecher

, Fehr

. Compassionate love for close others and humanity. J Social Pers Relat, 2005; 22:629–651.

33.

A. L. ESIZEREG: Stata module for calculating effect size based on a linear regression coefficient. Statistical Software Components S458607, Boston College Department of Economics, 2019.

34.

Willis

Patients' self-efficacy within online health communities: Facilitating chronic disease self-management behaviors through peer education. Health Commun, 2016; 31:299–307.

35.

Willis

, Royne

. Online health communities and chronic disease self-management. Health Commun, 2017; 32:269–278.

36.

Rupert

, Gard Read

, Amoozegar

, et al. Peer-generated health information: The role of online communities in patient and caregiver health decisions. J Health Commun, 2016; 21:1187–1197.

37.

Wang

, Kraut

, Levine

To stay or leave?: The relationship of emotional and informational support to commitment in online health support groups. In: ACM 2012 Conference on Computer Supported Cooperative Work, Seattle, WA, 2012, 833–842.

38.

Wang

, Zhao

, Street

. Analyzing and predicting user participations in online health communities: A social support perspective. J Med Internet Res, 2017; 19:e130.

39.

Myneni

, Cobb

, Cohen

. In pursuit of theoretical ground in behavior change support systems: Analysis of peer-to-peer communication in a health-related online community. J Med Internet Res, 2016; 18:e28.

40.

Balatsoukas

, Kennedy

, Buchan

, et al. The role of social network technologies in online health promotion: A narrative review of theoretical and empirical factors influencing intervention effectiveness. J Med Internet Res, 2015; 17:e141.

41.

Cohen

, Janicki-Deverts

. Who's stressed? Distributions of psychological stress in the United States in probability samples from 1983, 2006, and 2009. J Appl Social Psychol, 2012; 42:1320–1334.

42.

Fredrickson

, Boulton

, Firestine

, et al. Positive emotion correlates of meditation practice: A comparison of mindfulness meditation and loving-kindness meditation. Mindfulness (N Y), 2017; 8:1623–1633.