Self-administered mindfulness interventions reduce stress in a large, randomized controlled multi-site study

Mindfulness witnessed a substantial popularity surge in the past decade, especially as digitally self-administered interventions became available at relatively low costs. Yet, it is uncertain whether they effectively help reduce stress. In a preregistered (OSF 10.17605/OSF.IO/UF4JZ; retrospective registration at ClinicalTrials.gov NCT06308744) multi-site study (nsites = 37, nparticipants = 2,239, 70.4% women, Mage = 22.4, s.d.age = 10.1, all fluent English speakers), we experimentally tested whether four single, standalone mindfulness exercises effectively reduced stress, using Bayesian mixed-effects models. All exercises proved to be more efficacious than the active control. We observed a mean difference of 0.27 (d = −0.56; 95% confidence interval, −0.43 to −0.69) between the control condition (M = 1.95, s.d. = 0.50) and the condition with the largest stress reduction (body scan: M = 1.68, s.d. = 0.46). Our findings suggest that mindfulness may be beneficial for reducing self-reported short-term stress for English speakers from higher-income countries.

We determined the gender of participants based on self-reporting methods.Participants could answer to which gender they identified the most, being given six different options (i.e., male, female, transgender male, transgender female, prefer not to say, and an open answer in which they could write their gender).We did not collect disaggregated sex and gender data.We did not conduct sex-and gender-based analyses because the literature we had reviewed did not provide us with evidence to predict gender differences regarding mindfulness and stress reduction.
Reporting on race, ethnicity, or other socially relevant groupings Participants reported their country of origin and the country where they currently lived.This information was only used to describe the sample; we have not conducted any analysis involving such information.

Population characteristics
See above.

Recruitment
The sites involved in this project recruited participants with a Qualtrics link that was provided to them by the main investigator.The sites' coordinators were told by the main investigator that participants could be recruited using the SONA system of their respective institution or via crowdsourcing platforms such as mTurk or Prolific Academic.While using a combination of unpaid (e.g., SONA) and paid (e.g., Prolific) participation platforms could have mitigated self-selection bias, conducting the experiment solely online still limited our ability to completely eliminate this bias.Just like with practically any randomized controlled trial (RCT) on human subjects, self-selection of participants is inevitable.Participants, patients, etc., can decide whether they want to take part in the study.They can do so before and at any time during the experiment.Selfselection is critical when the goal is to describe a population (e.g., prevalence studies).It is, however, not a threat to the integrity of the results when the goal is to establish causal knowledge because, by definition, causal inference in RCTs is comparative, where we want to examine evidence for relative treatment effectiveness (Msaouel et al., 2023).The goal of an RCT is thus not to arrive at particular statements about the current state of the population, but rather identify and disentangle causal mechanisms.These principles are then likely transportable to the members of the given population, and frequently even beyond (Bradburn et al., 2020).

Ethics oversight
The study first received ethical approval from Swansea University's School of Psychology Research Ethics Sub-committee, while the sites that participated in the data collection either received ethical approval from their local IRBs or stated that they were exempt (e.g., if their IRB accepted the ethics approval awarded by Swansea University and did not request the local collaborator to submit their own application).Each site's IRB protocols with ethics details and acceptance of each protocol can be found on the OSF project page at https://osf.io/6w2zm/.Swansea University and Université Grenoble Alpes carried the administrative organization for the study.Swansea University was also the data controller for this project.The personal data of participants were processed for the purposes outlined in the information sheet (see the document Information Sheet at https://osf.io/xuznc/).Standard ethical procedures involved participants providing their consent to participate in this study by completing the consent form that was administered at the beginning of the online survey used for the experiment.
Note that full information on the approval of the study protocol must also be provided in the manuscript.

Field-specific reporting
Please select the one below that is the best fit for your research.If you are not sure, read the appropriate sections before making your selection.

Life sciences
Behavioural & social sciences Ecological, evolutionary & environmental sciences For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf Behavioural & social sciences study design All studies must disclose on these points even when the disclosure is negative.

Study description
The current multi-site project followed the route of a parallel randomized controlled trial.

Research sample
The study included participants from Australia, Europe, the UK, Canada, and the US.We had three exclusion criteria: 1) Participants had to be current non-meditators or to have not meditated in the 6 months prior to the experiment, 2) Participants had to be fluent or native English speakers, and 3) Participants had to declare they had not had a history of mental illness.Criterion 1 was used because the experiment focused on the effects of single brief exercises on non-meditators to better understand the potential benefits of mindfulness practices for this population.Criterion 2 was used because the audio files used were recorded in English.Criterion 3 was used because previous research has shown that mindfulness interventions have at times resulted in psychotic episodes, panic attacks, and depersonalization; thus, we needed to screen out participants for whom the mindfulness intervention could have been detrimental.

nature portfolio | reporting summary
April 2023 The final sample was not representative.After excluding participants that did not fit our inclusion criteria, we retained 2,239 valid observations (of these, 611 self-identified as males, 1,576 as females, seven as transgender males, two as transgender females, 27 did not identify with any choice, 16 preferred not to say; Mage = 22.4,SDage = 10.1;range 17-87; 94.2% students).
The rationale behind selecting the study sample (i.e., participants who were non-meditators, fluent in English and had no history of mental illness) was to specifically examine the effects of brief self-administered mindfulness interventions in isolation from potential confounds.A previous experience of engaging with meditation could have changed the baseline for any potential mindfulness effects; non-fluent English language proficiency could have created problems with mindfulness instruction comprehension; and a history of mental illnesses might have exacerbated potential meditation-related adverse effects (Britton, Lindahl, Cooper, Canby & Palitsky, 2021).This study aimed to deepen our understanding of how mindfulness practices can benefit individuals who are new to mindfulness.Although the sample is not representative of the general population, it provides valuable insights into the unique impacts of mindfulness on stress reduction measured experimentally, and practical relevance of mindfulness techniques among a predominantly young, student demographic spread across various geographic regions.

Sampling strategy
Each site collected the data using Qualtrics that redirected participants to the same survey; however, the URL address was tailored for each data-collecting site to allow recording the site participants belonged to.Participants were randomly allocated to the experimental conditions or control using Qualtrics' random block function.Prior to data collection, we conducted a simulation based on a Bayes Factor Design Analysis (BFDA) to assess the expected efficiency and informativeness of our study design.The simulation aimed to determine (1) the expected likelihood of the study to provide compelling evidence either in favor of H0 (BF01 = 1/10) or H1 (BF10 = 10), ( 2) the likelihood of obtaining convincing but misleading evidence, and (3) the likelihood that the study points in the correct direction even if stopped earlier due to pragmatic constraints on sample size.We modeled a sequential design with a maximum N of 720 and a minimum sample size of 420 participants per condition, with a goal of detecting an effect size of d = 0.20.We tested four interventions using a between-participants adaptive group design and found that the probability of arriving at compelling evidence was .79.A more detailed explanation of this simulation can be found at p. 10 of the manuscript.

Data collection
The experiment was conducted entirely online, and participants were instructed to complete it in a quiet environment for 20 minutes.Participants were asked to access the experiment using a desktop or laptop computer and not a mobile smartphone because one condition of the experimental design involved mindful walking, so we needed to ensure that any participant would be able to complete that task, if they were (randomly) distributed in that condition.The researchers were blind to the experimental conditions the participants were allocated to because the allocation was done using Qualtrics' randomizer function.

Timing
The data collection started on March 23rd, 2022 and finished on June 30th, 2022.

Data exclusions
The dataset originally comprised 6,691 responses, including both the 'test answers' generated by the site collaborators while developing and previewing the survey, and the actual answers submitted by the participants.From the survey's initial participants, we excluded the following: 1,307 who self-identified as meditators or reported having engaged in meditation within six months prior to the experiment, 776 who did not meet the English language proficiency requirement, and 981 who disclosed having a history of mental illnesses.Finally, 1660 participants started the survey without using a smartphone with headphones attached.Among these participants who failed to meet the inclusion criteria, 1,491 simultaneously met multiple exclusion criteria.Respondents who did not meet one or more inclusion criteria (N = 3, 233) were immediately directed towards the end of the survey, and we did not record further data from them.We also removed from analyses those who initiated the survey but did not progress up to the listening of the audio track (N = 976), and the 'test answers' provided by the collaborating researchers while developing the survey (N = 19); thus, the sample size dropped to N = 2,463.We then removed data from 19 participants that dropped out of the experiment and data from 205 participants who, according to our criteria, were considered careless respondents, yielding a final sample of 2,239 valid observations.Of these, 611 participants self-identified as male, 1,576 as female, seven as transgender male, two as transgender female, 27 did not identify with any choice, and 16 preferred not to say (Mage = 22.4,SDage = 10.1;range 17-87; 94.2% students), with an approximately even distribution across the five experimental conditions (Nmindful walking = 416, Nmindful breathing = 469, Nloving -kindness = 427, Nbody scan = 449, Nbook chapter -control = 478).We are not aware of how many participants were invited to the survey, but declined to participate.

Non-participation
We did not collect data from participants who declined to provide consent, as they were led to the end of the survey.However, participants who began the experiment but dropped out before the set of responses related to the main dependent variable were categorized as "careless participants" and were excluded from the main analyses.We do not have information regarding the reasons why participants may have abandoned the experiment.

Randomization
Participants were randomized to one of the experimental conditions (1,2,3,4) or to one of the active control conditions (story a,b,c).
As an example involving 15 participants,this is how they were expected to be randomized by the Reporting for specific materials, systems and methods We require information from authors about some types of materials, experimental systems and methods used in many studies.Here, indicate whether each material, system or method listed is relevant to your study.If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.