Effects of mindfulness meditation on anxiety, depression, stress, and mindfulness in nursing students: a meta-analysis and trial sequential analysis of randomized controlled trials

We searched seven English databases and three Chinese databases, including PubMed, Embase, Web of Science, Medline, PsycINFO, Cumulative Index to Nursing and Allied Health (CINAHL), Cochrane Central Register of Controlled Trials (CENTRAL), China Biology Medicine (CBM), Chinese National Knowledge Infrastructure (CNKI), and Wanfang. Databases were searched among literature collected from inception through August 2018 for randomized controlled trials (RCTs) in accordance with inclusion criteria. We developed a search strategy by PubMed on the basis of medical subject heading (MeSH) terms and text words of key articles which we identified a priori. We used a similar strategy in the other electronic sources. Search terms were “mindfulness”, “meditation”, “mindfulness-based meditation”, “mind–body therapies”, “mindfulness-based stress reduction”, “MBSR”, “Vipassana”, “Zen”, “pupil nurs*”, “nursing student*”. Studies were restricted to English and Chinese languages. Two reviewers searched articles independently according to search strategy.

Inclusion criteria were (1) the studies had to relate to MM directed only on nursing students; their nationality and race were not limited; (2) to assess anxiety, depression, stress, or mindfulness as outcome; (3) study design was RCT; (4) comparative control group; and (5) the studies were written in English or Chinese languages.

Exclusion criteria were (1) articles were not acquired; (2) articles without complete data; (3) articles about clinical nurses, medical students, or nursing teachers; and (4) duplicate literature.

Two reviewers screened titles and abstracts of relevant studies independently using search criteria. After full texts of these potentially included studies were obtained, two reviewers evaluated and selected them independently according to inclusion criteria. Disagreements between reviewers were resolved through discussions. The third reviewer was consulted for a final decision if consensus was not reached.

The methodological quality of the included RCTs was evaluated independently by two reviewers according to Cochrane Handbook for Systematic Reviews of Interventions.³⁰ The methodological quality of the included study in terms of a random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessors, incomplete outcome data, selective outcome reporting, and other bias exists. Each component was categorized as having a high, low, or unclear risk of bias. Disagreement was resolved by consultation with a third researcher.

Data were extracted using a standard data extraction form. Information such as study characteristics (e.g., first author and publication year), subjects characteristics (e.g., age and number of sample), study design, countries, method of intervention and control group, intervention duration, and outcome measures were obtained. Data were extracted by one reviewer and double-checked for accuracy by a second reviewer.

Meta-analysis was performed by RevMan 5.3 software (available from the website for free: http://ims.cochrane.org/revman). Standardized mean difference (SMD) was calculated for continuous outcomes with different measurements. We calculated SMD with 95% confidence intervals (CIs) for continuous outcomes. The heterogeneity was determined by chi-squared test and I² value (I² > 50% indicating high heterogeneity). A random effect model was used to estimate the overall effect when I² value ≥50%; otherwise, a fixed effect model was applied. Subgroup analyses used the random effect model. P < 0.05 was defined as significance level and P values were two-sided. Sensitivity analysis was conducted by eliminating every single trial to confirm the stability and reliability of the meta-analysis result. Due to the limited number of included studies, funnel plots were not conducted to assess publication bias.

Meta-analysis might result in type I errors due to an increased risk of random error when smaller numbers of RCTs and participants are involved. Thus, we undertook trial sequential analysis (TSA) to avoid random errors and examine the reliability and conclusiveness of the currently available evidence. If cumulative Z-curve was obtained across the monitoring boundaries, a sufficient level of evidence for the intervention effect may have been reached and no further studies were needed. To perform TSA, both types I and II errors were needed. In this review, the required information size (RIS) was estimated using α (type I error) of 5% two-sided and a β of 20% (power = 80%). The TSA was performed using software TSA 0.9 (available from the website for free: http://www.ctu.dk/tsa).

A total of 720 articles, including 544 English and 176 Chinese articles and another 2 from citation reference lists, were obtained from all the databases in preliminary search. After removal of duplicates, 502 titles and abstracts were examined for inclusion, and 27 full-text papers were identified and retrieved for detailed evaluation. After screening, 22 papers were excluded because they did not meet the inclusion criteria and 5 RCTs were included in the meta-analysis. The process of literature screening is shown in a flow diagram (Figure 1).

Figure 1

The characteristics of the included RCTs are summarized in Table 1. Total of 5 RCTs involving 257 subjects were included and all reports were written in English language. The age of the participants was not recorded in one study. One study was conducted in the USA,³¹ two in Korea,^32,33 one in Thai, and one in China.^34,35 The sample of trials ranged from 41 to 60. The mode of intervention was face to face with the participants in three studies,^33,34,35 a combination of voice-recorded files and face to face in one study,³² and online instruction in one study.³¹ The duration of intervention was 1 week with 7 sessions in one study,³⁵ 4 weeks in two studies^31,34 and 8 weeks with 8 sessions in two studies.^32,33 A total of four studies assessed anxiety. Anxiety was measured using the Self-Rating Anxiety Scale (SAS) in one study,³⁵ the State Trait Anxiety Inventory (STAI) in two studies,^32,34 the Depression, Anxiety and Stress Scale-Anxiety (DASS-A) in one study.³³ Three studies assessed depression, one study used the Beck Depression Inventory,³² one used the Self-Rating Depression Scale (SDS),³⁵ and the remaining one study used a Depression, Anxiety and Stress Scale-Depression (DASS-D).³³ Four studies assessed stress, two studies used the Perceived Stress Scale (PSS),^31,34 one used the Psychosocial Wellbeing Index-Short Form Questionnaire (PWI-SF), and one study used a Depression, Anxiety and Stress Scale-Stress (DASS-S).^32,33 Two studies measured mindfulness by the Mindfulness Attention Awareness Scale (MAAS) and the Five-Facet Mindfulness Questionnaire (FFMQ), respectively^31,33 and two studies used PSS.^31,34 All studies were published between 2009 and 2017.

Risk of bias assessment is presented in Figure 2. Of all included RCTs, two studies were grouped by computer-generated random number table and each participant drew a number, respectively,^32,35 while remaining three failed to report details.^31,33,34 Two trials reported low risk on allocation concealment.^32,35 None of the included RCTs described blinding of the participants and personnel. The outcomes assessor was reported to be blinded in two studies.^32,35 In assessment of attrition bias, all studies were believed low risk because the pre-specified expected outcome of the studies was clearly reported. All studies reported the results to avoid reporting bias. In addition, all trials had described that there were no statistical difference in baseline characteristics between the groups.

Figure 2

To evaluate the effect of an individual study on the results for the meta-analysis of anxiety, depression, stress, and mindfulness, sensitivity analysis was performed. Results of the effect on depression and mindfulness were changed when studies were eliminated. Excluding the study results by Chen and Song,^33,35 respectively, results of the effect on the level of depression and mindfulness were changed, which means the conclusion is unreliable. More studies with a larger sample size will be needed for the results to be verified.

An increasing body of evidence has demonstrated that MM can also significantly improve physical and mental symptoms for different groups of people, such as breast cancer survivors, elementary school teachers, chronic insomnia patients, and so on.^36,37,38 The study showed that the psychological pain of the students was reduced and their quality of life was improved after attending MM training.³⁹ Also, meditation training has a positive effect on the awareness and cognition of healthy people (perception, attention, memory, etc.) which can improve attention and cognitive flexibility.⁴⁰

This meta-analysis provided evidence supporting the effects of MM in nursing students. To achieve high quality rather than quantity, our meta-analysis included only five RCTs, all of which were published in recent 10 years. This reflects the growing interest for MM in the nursing education. We carried out TSA to determine the RIS and to evaluate whether the current evidence is sufficient. The study found that TSA could protect against spurious conclusions from meta-analysis, especially from those that reported positive results based on only few studies and few events.⁴¹ Another study showed that TSA could prevent 93% of false-positive results.⁴²

The meta-analysis systematically reviewed currently available literature, including 5 RCTs with 257 participants to evaluate the effect of MM on anxiety, depression, stress, and mindfulness in nursing students. The pooled results of the involved RCTs revealed that MM significantly reduced the level of anxiety and stress and the results manifested reliability with current evidences according to TSA. We observed significant improvements in the control of anxiety, stress, and mindfulness which support implementing MM programs for nursing students.

In addition, the results of our subgroup analyses of the depression suggested that intervention duration of 8 weeks had positive effects but short intervention duration (1 week) had no significant effects. These results can be explained for several reasons. First, depression is due to long-term accumulation of psychological problems which needs to be improved by a step-by-step process. Only after a certain intervention time, its positive effect may be revealed.³⁵ Second, included the number of studies is insufficient, there may be a publication bias.

However, there was no significant effect on increasing mindfulness level and we could not make definitive conclusions with current evidences according to TSA. Therefore, researchers should conduct a large number of high-quality RCTs in the future to explore the impact of meditation on the level of depression and mindfulness in nursing students in order to give full play to the positive role of evidence-based nursing theory.

Some strengths of our study must be pointed out. First, we performed a comprehensive database search and identified all relevant RCTs as many as possible. Second, we conducted a rigorous quality assessment and most of the trials included are of moderate quality. Furthermore, although there are some methodological differences between the studies, most of the statistical heterogeneity was low or absent in the analyses. Finally, we also performed detailed analyses of the data and through TSA we could confirm a significant effect that MM can improve anxiety and stress for nursing students.

Some limitations of this meta-analysis must be acknowledged. First, the overall sample size was not large enough, so the result may be resulted in a certain bias. Second, included outcomes were measured using different self-reported scales, so selection bias may be caused. Further studies in this field should use the same assessment instruments, so that other researchers can conduct systematic reviews enrolled a larger sample size. Third, articles published only in English or Chinese languages were selected due to language constrains. This may cause a potential risk of publication bias.

The MM program, a non-pharmacological approach to ease bad mood, emphasized on self-care, compassion, and healing, makes it an effective intervention for stress management.⁴³ Mindfulness is the same as the present consciousness, and seems to be positive psychology or spirit. This concept is similar to resilience, which is the ability to transform from a negative experience to a positive way in physiology and psychology.⁴⁴ Since nursing students may experience high levels of stress in their educational environment which lead them to a state of anxiety or depression, it is critical for nurse educators that students’ health is regard as a priority. Educators can help students by integrating MM intervention programs into the educational curriculum which target management of anxiety, depression, and stress. Higher mindfulness scores of meditations may also have implications for safe nursing practice, as the mindfulness trait carries with it a heightened monitoring ability and meta cognitive skill.⁴⁵ MM can provide nursing students with a simple, effective way to cope with anxiety, depression, and stress, increase mindfulness level, reduce bad emotion elicited physical arousal, and achieve mental well-being and professional growth in their future careers. So MM program is worthy of application among nursing students.