The sample was comprised of 118 participants (intervention group, 60; waitlist control group, 58) recruited from the general population. With an age range of 21–62 years (M = 40.68, SD = 10.45), 95 participants (80.5%) were female. The majority of participants (94.9%; n = 112) worked full-time for a mean of 45.12 (SD = 14.84) hours/week and were married or had a partner (n = 85; 72%), with 59 (50%) having dependent children. One hundred and two participants (86.4%) worked a 9 am–5 pm (Mon-Fri) pattern, with the remaining participants working shifts. Many job roles were represented including: nursing/medicine (26.3%; n = 31), healthcare (e.g. dieticians, physiotherapists; 20.3%; n = 24), administration (19.5%; n = 23), education (e.g. teachers, University lecturers; 14.4%; n = 17), management (8.5%; n = 10), police (6.8%; n = 8) and other (4.2%; n = 5). Roughly two thirds of participants were University educated (68.6%; n = 81).
Approximately 50% of the sample reported moderate to severe levels of depression and/or anxiety symptoms at baseline (see Table 1); however, only five participants self-identified as having depression or anxiety (three from the intervention group, and two from the waitlist control group), and two of these participants stated they were taking medication. Participants attracted to taking part in this study may have been seeking help due to the severity of experienced symptoms. As part of the consenting process, participants agreed to complete the course within 4 weeks if possible. See Fig. 1 for the participant flow from screening to follow-up, and Table 1 for sample specifics for each of the study groups.
A randomised waitlist control design was employed. Participants were assessed pre-treatment and post-treatment and were followed up at 3 and 6 months post-treatment.
Details of the study were circulated to organisations within the UK with which the University had relationship in order to promote the study to their staff. For reasons of confidentiality, the specific organisations cannot be named but they span the following industry sectors: Healthcare (e.g. nursing, medicine), Policing, Legal, Education, Information Technology and Telecommunications. In addition, the study was promoted via social media and was also advertised on an online professional networking site (www.LinkedIn.co.uk).
Individuals completed an online screening questionnaire. To be eligible for inclusion, participants had to (1) be 18 years of age or older; (2) be working a minimum of 30 h per week; (3) have the ability to commit to 2 h (minimum) per week for the duration of the course; (4) have access to the Internet at home; (5) not be receiving any other form of psychological therapy and no plans to start any other form of therapy throughout the duration of the study; (6) have no previous experience of mindfulness or meditation; (7) agree to maintain any dosage of existing medication during the study, but in the event that dosage needs to change during the study for medical reasons to notify study personnel; and (8) be living and working in the UK.
Randomisation did not occur until all participants had registered for the study. Allocation concealment was achieved by allocating unique identifiers to each participant and then randomly sorting the file in SPSS version 21 (IBM Corp 2012). Participants, using the unique identifiers, were then randomly assigned into blocks of four (stratified by gender) which were generated using a random number generator program (Urbaniak and Plous 2013). Allocation to even numbers in the block denoted intervention group membership, and to odd numbers denoted waitlist control group membership. We stratified by gender because previous research has highlighted gender differences regarding the prevalence and severity of anxiety and depression (McLean et al. 2011). Participants were blinded to group membership. They were not able to choose which group they were allocated to however they were informed there were two course start dates. Participants had no contact with each other because all recruitment was conducted online, and all communication with participants was conducted via personal email. The data set used for analysis contained only an anonymised participant unique identifier for each participant.
Compensation for Participation
Participants were offered £50 worth of Love2Shop vouchers to compensate them for their time and for expenses associated with completing the course. In addition, participants were informed that they were completing an online course for free which would normally cost them £30.
Online Mindfulness Course
The online mindfulness-based cognitive therapy (MBCT; Teasdale et al., 2000) course (www.bemindfulonline.com) is run by the Mental Health Foundation (UK) and Wellmind Media (UK), and was developed in conjunction with leading UK mindfulness instructors (Krusche et al. 2012). The course usually costs £30 per person; however, participants in this study were able to complete the course for free. The course is modelled on the class sequence of traditional MBCT programmes. In the current study, participants were asked to complete the course within 4 weeks if possible. Participant completion was tracked throughout the course. Participants were sent reminder emails when they had not accessed the course for more than a week. Participants did not have any personal contact with the mindfulness instructors at any point during the course. All instructional video and audio files were embedded within the website. For further detail regarding the course, see Krusche et al. (2012) and Querstret et al. (2016).
The Perceived Stress Scale 10 item (PSS-10; Cohen et al. 1983) consists of 10 items (e.g. “In the last month, how often have you felt that you were unable to control the important things in your life?”) which are answered using a five-point Likert scale, ranging from 0 (Never) to 4 (Very often). A total score is computed by summing the scores on the individual items with scores ranging from 0 to 40. Cronbach’s alphas: T1 = 0.86, T2 = 0.88, T3 = 0.91, T4 = 0.92.
The Patient Health Questionnaire 9-item (PHQ-9; Kroenke et al. 2001). Participants are asked to consider over the last 2 weeks, how often they have been bothered by a list of nine problems, for example, “Little interest or pleasure in doing things”. Items are scored against a Likert scale ranging from 0 (Not at all) to 3 (Nearly every day) with summed today scores ranging from 0 to 27. Depression severity is determined on the basis of the total score, as follows, 0–4 = No depression; 5–9 = Mild depression; 10–14 = Moderate depression; 15–19 = Moderately severe depression; and 20–27 = Severe depression. Cronbach’s alphas: T1 = 0.88, T2 = 0.89, T3 = 0.89, T4 = 0.87.
The Generalised Anxiety Disorder 7-item (GAD-7; Spitzer et al. 2006). Participants are asked to consider over the last 2 weeks, how often they have been bothered by a list of seven problems, for example, “Feeling nervous, anxious, or on edge”. Items are scored against a Likert scale ranging from 0 (Not at all) to 3 (Nearly every day) with summed total scores ranging from 0 to 21. Anxiety severity is determined on the basis of the total score, as follows, 0–4 = No anxiety; 5–9 = Mild anxiety; 10–14 = Moderate anxiety; and 15–21 = Severe anxiety. Cronbach’s alphas: T1 = 0.92, T2 = 0.92, T3 = 0.92, T4 = 0.90.
The Five Facet Mindfulness Questionnaire Short form (FFMQ-SF; Bohlmeijer et al. 2011) has 24-items that measure five facets of mindfulness: observing (OBS; 4 items, e.g. I notice the smells and aromas of things), describing (DES; 5 items, e.g. I’m good at finding the words to describe my feelings), acting with awareness (AA; 5 items, e.g. It seems I am “running on automatic” without much awareness of what I am doing), non-judging (NJ; 5 items, e.g. I criticise myself for having irrational or inappropriate emotions) and non-reactivity (NR; 5 items, e.g. I watch my feelings without getting lost in them). Participants are asked to rate the degree to which each statement is true for them. Items were scored on a five-point Likert scale ranging from 1 (never or very rarely true) to 5 (often or always true), with summed facet scores ranging from 5 to 25. Previous research has shown that the OBS facet is only predictive for participants with previous experience of meditation (Baer et al. 2006), and the participants in the current study were required to be naive to meditation; therefore, this facet was not utilised. Cronbach’s alphas: DES (T1 = 0.84, T2 = 0.85); AA (T1 = 0.79, T2 = 0.86); NJ (T1 = 0.78, T2 = 0.87); NR (T1 = 0.82, T2 = 0.83). We did not assess facets at 3- and 6-month follow-up because they were included as mediators, and the mediation models could not be tested beyond post-treatment due to the waitlist control group commencing the intervention.
Sample Size Calculation
Lakens and Evers (2014) propose that in order to find a medium effect size between two groups, with power of 0.80, 41 participants are required in each group (Total n = 82). An a priori power analysis for an analysis of covariance (ANCOVA) computed using G*Power 3.1.9 (Faul et al. 2009) determined that a target sample size of 90 participants was required to sufficiently power the study at a 0.80 level to find a medium effect size.
Step 1: In service of our main study hypotheses (H1, H2, H3), we assessed the effect of the intervention on perceived stress, depression and anxiety immediately after intervention group participants had completed the course (end of the waitlist period for waitlist control participants). Step 2: To identify mindfulness facets to be included as mediators in subsequent analyses, we conducted a manipulation check to see if the intervention had affected one, some, or all of the facets. For steps 1 and 2, data were analysed using multivariate analysis of covariance (MANCOVA) and univariate ANCOVA in SPSS version 21 (IBM Corp 2012). Step 3: Using the PROCESS macro (Hayes 2013), we assessed the mindfulness facets affected by the intervention in a multiple mediation model. In our bootstrap analysis, we specified 10,000 resamples and 95% confidence intervals with confidence intervals including zero indicating a null effect (Mooney and Duval 1993). Step 4: Using repeated measures ANOVA, we assessed whether the effects of the intervention were maintained at 3-month and 6-month follow-up (H4). Step 5: We assessed clinically significant change using chi square tests for depression and anxiety because at baseline almost half of the sample were moderately to severely affected.
Multiple Imputation for Missing Data
Fifteen participants in the intervention group did not complete the intervention, and were recorded as drop-outs. All waitlist control participants completed the waitlist period. The dropout rate in the current study (25%; see Fig. 1) was comparable to other studies (for review, see Swift and Greenberg 2012). Best practice dictates that Intention-To-Treat (ITT) principles be adopted; whereby, all randomised participants are included in the analysis in their allocated groups, irrespective of treatment adherence or completion (Altman 2009). In the current study, missing data was imputed (five iterations) using the multiple imputation process in SPSS version 21 (IBM Corp 2012). In the multiple imputation model, in order to provide a good prediction of missing values, the following variables were entered: baseline (T1) scores for perceived stress, depression and anxiety; T1 scores for AA, DES, NJ, NR and demographic variables (age, gender, children [yes/no], time in current role [years], job status [full-time/part-time], job pattern [traditional/shifts], hours worked per week, level of education). We report ITT and Per Protocol (PP) results throughout in all Tables. We interpret our findings in respect of the ITT results, and only include reference to PP results in text where there is a significant difference.
Prior to conducting the main analyses, correlation analysis was carried out on the main study variables and mindfulness variables in order to test the MANCOVA assumption that the main study variables would be correlated with each other—and that the mindfulness variables would be correlated with each other (Meyers et al. 2013). A meaningful pattern of correlations was observed amongst the main study variables, and also amongst the mindfulness variables (see Table 2).