Virtual reality hypnotherapy for healthcare students: a feasibility trial

Abstract

Purpose

There is evidence that both hypnotherapy and virtual reality (VR) can be helpful in reducing perceived stress in the general population. This is a feasibility and acceptability trial of an intervention combining hypnotherapy and VR to establish its acceptability in students. This study aims to establish whether students found the experience acceptable, described any adverse effects and whether they reported feeling calmer after the experience.

Design/methodology/approach

The study was testing the hypothesis that students would attend the sessions and find the experience acceptable. A secondary hypothesis was that preliminary qualitative and quantitative evaluation of measures of stress and wellbeing would signal potential improvements.

Findings

All participants completed all three sessions. No side effects were reported. Visual analogue scales recorded each day assessing the immediate effect improved. At the end of the intervention, there was an increase in wellbeing of 2.40 (95% CI: 1.33, 3.53, p = 0.006), and a decrease in depression of 0.73 (95% CI: 0.40, 1.07, p = 0.010), reflecting large effect sizes of 0.76 and 0.83, respectively. Qualitative feedback was generally very positive.

Research limitations/implications

This study is small with just 15 students and was over a short period of time. The recruitment method meant there was no way to establish whether the volunteer students were representative of the general student population in terms of mental wellbeing. There was no control arm.

Practical implications

The preliminary results suggest that a larger controlled trial is justified.

Social implications

This VR experience may have benefit to university students and to the wider population.

Originality/value

This described the evaluation of a novel intervention for perceived stress combining hypnotherapy and virtual reality in a group of healthcare students, with promising results suggesting further evaluation is needed.

Citation

O’Brien, A., Hutchinson, J., Bin Fauzi, N., Abbott, M., Railton, J., Bell, D., White, S., Smith, J. and Riches, S. (2024), "Virtual reality hypnotherapy for healthcare students: a feasibility trial", Mental Health and Digital Technologies, Vol. 1 No. 2, pp. 160-172. https://doi.org/10.1108/MHDT-01-2024-0005

Background

There has been a trend of increased reporting of mental health conditions by students entering higher education; for full-time undergraduate entrants, this has increased from 0.7% in 2010–11 to 4.5% in 2021–22 (Office for Students, 2023). Universities have documented a corresponding rise in demand for counselling services (Thorley, 2017). Students studying clinical subjects such as medicine are under even greater pressure than those on non-clinical courses (MacLean et al., 2016; Rotenstein et al., 2016; Student BMJ, 2014). This seems to have worsened with the Covid-19 pandemic; in one study, 55% of medical students reported moderate to extreme levels of stress (O’Byrne et al., 2021). Student mental health is, therefore, a national priority.

Universities are increasingly looking for innovative, evidence-based ways to support students and their welfare needs. Students find it difficult to engage with existing interventions and often do not seek help (Thorley, 2017). Emerging literature, however, suggests that they may engage particularly well with interactive technologies (Schindler et al., 2017) and show a high degree of acceptance of virtual reality (VR) (Cabero-Almenara et al., 2023). A systematic review by Xu et al. suggests that VR is an accessible medium for stress reduction for students and young adults (Xu et al., 2024) and VR has been described as safe, engaging and beneficial for medical students (Dhar et al., 2021).

Hypnotherapy for wellbeing

Hypnotherapy creates a state of focused attention and increased suggestibility. There is an evidence base supporting the use of hypnotherapy techniques for the management of perceived stress and anxiety (Fuhr et al., 2021; Williamson, 2019; Chen et al., 2017; Häuser et al., 2016; Jensen et al., 2015). A systematic review in 2009 suggested that hypnotherapy may be effective in reducing exam anxiety in medical students; however, only five small randomised controlled trials were identified (Baker et al., 2009).

Virtual reality

Virtual reality technology creates an immersive and interactive setting (Freeman et al., 2017). VR headsets are devices where virtual environments are displayed in three dimensions using a head-mounted-display. VR technology has been used in patients with mental health problems to aid research into aetiological models and has been used in the potential treatment of various conditions including anxiety, phobias and post-traumatic stress disorder (Freeman et al., 2017). Some of the more established uses are in cognitive behaviour therapy (CBT), particularly exposure therapy for phobias (Parsons and Rizzo, 2008) and in relaxation techniques for generalised anxiety disorder (Carl et al., 2019; Kothgassner et al., 2019). Recent reviews concluded that VR may be a useful and promising tool to promote relaxation and reduce stress in both clinical and non-clinical samples (Riches et al., 2021). Another review suggests that VR has the potential to support mindfulness practice (Arpaia et al., 2021).

Hypnotherapy via virtual reality

Virtual reality hypnotherapy (VRH), where visual experiences are combined with audio hypnotherapy, is defined as “hypnotic induction […] delivered by customised virtual reality software” (Patterson et al., 2010).

There is little theoretical underpinning for the combination of VR and hypnotherapy in stress reduction but there has been discussion in the literature on pain modulation. It is postulated that pain reduction could be due to distraction enabled by the VR environment (Spiegel, 1991). Some authors believe that VRH could be more useful than traditional hypnotherapy because it easily focuses the attention (Askay et al., 2009).

Side effects and safety

VR headsets are not currently recommended for use in people with epilepsy or with a pacemaker. A relatively commonly documented side effect of VR is “cybersickness” (CS), which presents similarly to motion sickness. Symptoms of CS usually resolve within minutes of removing the VR headset (Askay et al., 2009) and can be minimised or avoided by keeping the participant stationary. CS is reduced with more modern devices (Caserman et al., 2021). There have been no other reported adverse effects.

Summary

This is a feasibility trial of a new VRH experience for student wellbeing. As a novel experience, there were uncertainties about whether students would find the experience acceptable or describe any adverse effects. The plan was to establish the acceptability of the intervention and aid the design of a larger controlled trial.

Feasibility of the intervention was assessed by reporting the intervention completion rates and any incidents of side effects. Potential efficacy was assessed by obtaining measures of stress, wellbeing, depression and anxiety. These were taken pre and post each session. Any cumulative effects over the three days were noted. Acceptability was evaluated by asking the participants about their experience of the intervention.

Methods

The study was conducted in July 2022. It was originally designed to run over five consecutive days but extreme heat weather warnings restricting travel led to it being shortened to three days.

Setting

St. George’s University of London (SGUL) is the only specialist health-care university in the UK. Located in southwest London, it serves 5,500 full time students on a range of courses: medicine, biomedical science, physiotherapy, diagnostic and therapeutic radiography, physician associate studies and paramedic science. In total, 60% percent of the students are female. Approximately 46% are White, 33% Asian, 11% are Black and 9% are other/mixed.

Participants

Students from all courses were invited to volunteer for the study via the student union and personal tutor network. All 5,500 students enrolled at the university therefore had the opportunity to take part as all received the emails from the student union. Students were asked not to volunteer if they had a history of seizures or a pacemaker.

There were no other exclusion criteria beyond being under the age of 18. No exclusions were made on the basis of mental health issues. The feasibility trial presented an opportunity to practically establish the functioning of the headsets, the appropriate physical environment and timing of the intervention (both time in the day and in the student term timetable.)

Intervention − the VR hypnotherapy experience

The process builds upon established elements of hypnosis; a slightly altered state of consciousness, when clear and explicit suggestions are made to participants. The experience was developed by an interdisciplinary team bringing together experience with VR production and design expertise. It was designed by VR producers in conjunction with a clinical hypnotherapist who has a history of working with medical students (UJ). UJ has run hypnotherapy courses for medical students for many years and it was in discussion with the VR producers that the intervention was developed. This pilot aimed to establish whether students found the experience acceptable.

The experience lasts approximately 6 min. UJ’s voice guides viewers, with suggestions regarding breathing and relaxation alongside sound and music. The audio is augmented with generative VR graphics beginning in a domed hall. The viewer is guided through a transition into darkness, then rings of particles form ahead of the viewer and approach, passing around them on all sides before fading. The timing of these rings as well as their pulses aid UJ’s suggestion to slow breathing down. UJ then guides the viewer to further relax before they are gradually brought back to the domed hall. The viewer is then grounded in the present and asked to remove their headset as the experience finishes.

Procedure

Students who volunteered for the feasibility trial, which was single arm, were allocated a project ID and their sex, age group and ethnicity recorded. Participants were asked to attend a quiet annex of the student union for a maximum of 1 h each day (timed for the end of the teaching day) on three consecutive days. At the start of the first session, participants completed the following self-report scales which are all commonly used tools with strong psychometric properties:

• Wellbeing – Warwick-Edinburgh mental wellbeing scale (WEMWBS) (Tennant et al., 2007), 14 items. Total score ranges from 14 to 70, with scores of 14−42 indicative of low wellbeing and score of 60−70 indicative of high wellbeing (Tennant et al., 2007).

• Perceived stress − Perceived stress scale (PSS) (Cohen et al., 1994), 10-item version. Total score ranges from 0 to 40; scores ranging from 0 to 13 are considered low stress, scores of 14–26 are considered moderate-perceived stress and scores between 27 and 40 are considered high-perceived stress.

• Depression – Patient health questionnaire-2 (PHQ2) (Kroenke et al., 2001), 2 items. Total score ranges from 0 to 6 with a score of 3 or greater a positive screen for a depressive disorder (Kroenke et al., 2001).

• Anxiety – Generalized anxiety disorder (GAD-7) questionnaire (Spitzer et al., 2006), 7 items. Total score ranges from 0 to 21 with mild, moderate and severe levels of anxiety indicated by scores of 5–9, 10–14 and 15–21, respectively (Spitzer et al., 2006).

The scales were all repeated at the end of the third session. Before and after each VR hypnotherapy (VRH) session, participants were asked to complete simple visual analogue scales asking how stressed, happy, sad, calm and anxious they were on a scale of 0 to 10. Pulse and blood pressure readings were also taken pre and post sessions to be assessed as proxy indicators of stress, by junior doctors in psychiatry (D.B., J.R., M.A. and N.B.F.). Pulse was measured manually and an electronic sphygmomanometer was used for blood pressure.

At the end of the study, participants were asked to complete a short paper semi-structured questionnaire with open text responses about their experience and acceptability; what was good, what was bad, what they think could have been improved about the experience and any further free text comments.

Students were reimbursed with a 30 pounds voucher.

Sample size

This study aimed to recruit 20 participants to be able to estimate the potential efficacy of the intervention. Twenty participants are sufficient to be able to detect a within-group effect size of 0.7 with 80% power at a 5% significance level. Summary statistics of the outcome measures, markers of perceived stress, well-being and affective function, will be used to inform sample sizes for future two-arm trials.

Data analysis

Quantitative analysis

Frequencies and percentages were used to describe the sample characteristics. For the quantitative data, initially a series of three (session) by two (intervention) repeated measures analysis of variance (RMANOVAs) models were computed to determine whether there was any effect of session or an interaction effect between session and intervention. No significant effects of session or session x intervention interaction were observed and therefore pre- and post-intervention values for each emotion visual analogue scale and biological markers were calculated by averaging the respective values obtained for each session. The resultant pre- and post-intervention values and those of the broader outcome measures (markers of perceived stress, well-being and affective function administered once prior to the intervention and again at the end of the VRH intervention period) were compared using paired-sample t-tests and Hedges’ g effect sizes obtained (Hedges, 1981). Where either of the paired data distributions failed to approximate a Gaussian distribution (i.e. skewness < −1 or > 1, kurtosis < −1.5 or > 1.5) (Hair et al., 1998), bootstrapping was used to calculate 95% confidence intervals of the mean differences and associated p-values, and effect sizes calculated with values transformed using Box-Cox methods.

Qualitative analysis

The open text responses from the brief semi-structured questionnaire were transcribed and entered into Excel by A.A.O’B. These were exported into NVivo and analysed using a conventional content analysis (Hsieh and Shannon, 2005) by J.H. This was in turn used to describe the experience of the participants and acceptability of VRH. The analysis was inductive and although the responses were limited by the questions asked, the codes were based on participant responses rather than being pre-conceived. Initial coding involved reading through the data set for familiarisation several times and then coding short segments descriptively. This was a recursive process, with continuous review by comparison between and within. This resulted in codes being merged or split, and then organised into a hierarchical structure with several superordinate and subordinate categories. The emerging codes and hierarchies were reviewed with the primary researcher (A.A.O’B.) to ensure trustworthiness. The coverage percentage indicates how much of the source content is coded.

Ethical approval was granted by the St. George’s Ethics Committee. Students were aware that if they felt uncomfortable at any time during the experience, they could remove the headset and were supervised during the experience by junior psychiatrists.

Results

Sample description

Twenty students replied to the invitation emails, and none had a history of epilepsy or a pacemaker. However, just before the study started, five students withdrew due to the travel restrictions imposed by the weather warning. No demographic data were collected on these students. All students completed all the scales.

Therefore, 15 participants entered the study. The sample was 80% female (12). Around 60% were of white ethnicity (nine), 20% Black (three), 13% Asian (two) and 7% mixed (one).

In total, 67% of the sample (ten) were aged between 21 and 25 years, 20% aged between 18 and 20 years (three) and 13% aged above 25 years (2).

Using established cutoff values, the sample can be described as: 40% (six) had low wellbeing, one participant reported high stress with 13 (86%) reporting medium stress.

Four participants (27%) screened positive for depression and five (33%) reported moderate anxiety with two (13%) reporting severe anxiety.

All participants (100%) completed all three sessions and all the questionnaires. No participant withdrew from any session. No side effects were reported. None reported cybersickness (they were advised to remove the headset if this was experienced).

Potential efficacy of VRH on perceived stress, well-being and affective function

The mean [standard deviation (SD)] values for measures of perceived stress, well-being, depression and anxiety before and after the three days of VRH sessions are shown in Table 1. There was an improvement in well-being at the end of the intervention period, an increase in WEMWBS of 2.40 (95% CI: 1.33, 3.53, p = 0.006), and a significant decrease in PHQ-2 scores of 0.73 (95% CI: 0.40, 1.07, p = 0.010), reflecting large effect sizes of 0.76 and 0.83, respectively. There was little change in perceived stress. Anxiety decreased over the three days, although the change was not significant, p = 0.078, with the effect size in the medium range.

With regard to the immediate effect of the VRH on the emotion visual analogue scales, it can be seen from Figures 1(a), 1(b) and 1(c), that ratings improve significantly across the five measures with effect sizes ranging from 0.82 to 1.96.

Mean participant pulse rate (per minute) and blood pressure pre- and post-session are shown for each session on Figures 2(a) and 2(b). The graph suggests a small but reliable decrease in pulse rate over the course of each session of VRH but a mixed picture on systolic and diastolic blood pressure. The mean pre- and post-session values, shown in Table 1, indicate a moderate but significant effect of the intervention on pulse rate, effect size = 0.56, but no significant impact on blood pressure indices.

Qualitative analysis

There were four categories suggested by the data. These were: positive experiences; impact and intention; set up and context; difficulties and improvements.

Positive experiences

Most of the participants described experiencing the trial positively (20.15% coverage). These positive experiences fell into two sub-categories. The first of these was experiencing guided breathing as supportive, and particularly the synchronicity of the breath and the imagery. For example, one participant describes how they found “the Kaleidoscope pattern moving in timing with breathing which helps with calming down” useful (participant 7). For many, it seemed that the combination of the audio and the visuals led to a more immersive experience than that of an audio-only meditation, and this sense of immersion was the second sub-category. As one participant noted “soothing voice and music alongside the moving light (is) much better than a regular audio guided meditation because of how immersive it is” (participant 1). This was not universal, and one participant found that the experience was not immersive enough.

Impact and intention

Most of the participants noted that there was a helpful short-term impact to each session which was in the main experienced as relaxing or calming, for example, participant (6) “did genuinely feel more relaxed and less anxious afterwards. Even if only for a short while”. This was a robust and common response (32.3% coverage) though other pleasant impacts were also noticed ranging from feeling energised to more focused, noticing improved mood and a different relationship to thinking. As one participant (10) reported:

My thoughts became easier to sort through and less stress was associated to those thoughts, felt more grounded to the present, and not thinking about what’s next or in over things already done.

There were a minority of participants who described some potentially unpleasant impacts such as feeling sleepy (1), and a few who described feeling unsettled by the experience (2). One participant described feeling unsettled “scared at first when looking down and not seeing my body and it would catch me off guard when the audio asks you to look at the light travelling beneath you” (3).

Set-up and context

How things were explained was helpful in settling people into the experience, for example “it was a peaceful environment. The VR session was good and very straightforward. I understood fully what was going on and was to immerse myself in the experience (10) and “everyone was friendly. I was always kept informed, great organisation. I felt safe” (5). Despite the majority of participants describing the friendly and organised set-up of the sessions, it still took a little time for some participants to settle into the experience and there were (n = 7) comments related to this. Most of the participants described increasing levels of comfort as the sessions progressed but there did seem to be a period of transition.

Difficulties and improvements

The fourth category included the kinds of difficulties participants experienced and what improvements they thought might help. The majority of participants suggested, in line with comments about the transition into the practice, that having longer sessions would be beneficial, for example “longer sessions more focus on getting comfortable initially” (13) although this was not universal. Other preferences mentioned were to have a choice of voice (sex), headset issues including managing these with glasses, some more variety in the guided practices offered, and some support with triggers or fears experienced.

Discussion

This was a study piloting a novel VRH experience, designed by independent experts in VR and in clinical hypnotherapy with an interest in health-care student stress management and piloted by staff involved in student welfare. All 15 students who took part completed the three days of the experience. Of note, there were no participants that reported sensations consistent with cybersickness.

The key feasibility finding was that the attending students were generally very positive about the experience and described no adverse effects. For a further larger trial, it was felt that more extensive communications would be needed to recruit participants; for example, via the weekly all-university student letter and via the communications department. The immediate positive effect of the VRH post session was shown across all visual analogue scales, although these ratings are very susceptible to bias given the context. Students’ mental wellbeing and depression scores signalled improvement over the three days.

Of note in this study was the qualitative enthusiastic response from the students. There were a few negative comments about the practicalities, describing issues about the headset design. However, most of the students described the experience as calming, positive and something they would like to use in the future. The feedback will be used to amend the experience in preparation for further evaluation in a randomised controlled trial.

Our findings support the growing research that suggests that VR is a feasible and acceptable tool to promote relaxation (Riches et al., 2021). Previous studies have established the mental health benefits of VR relaxation in keyworkers, such as health-care professionals, experiencing high levels of work-related stress (Riches and Smith, 2022) and there may be a range of applications for VR and immersive technologies in the workplace that could be applicable to students. This pilot study adds to the emerging literature focussing on VR in the student population (Modrego-Alarcon et al., 2021; Camara and Hicks, 2020). Liu et al. found that VR may enhance a mindfulness programme in students (Seabrook et al., 2020) and that VR seemed to encourage attendance on the programme. VR-enhanced dialectical behaviour therapy (DBT) was found to be more effective than DBT alone (Lamb et al., 2022).

The emerging evidence therefore suggests that VR interventions to support student wellbeing are well received and may be helpful. Given that there have been few studies exploring the use of VR generally and even fewer specifically in health-care students, the positive findings of this study add weight to the need for further investigation of the field.

Limitations

This study is small with just 15 students and was over a short period of time. However, the data does suggest that hypnotherapy via VR has potential use in students. The convenience recruitment sampling method meant there was no way to establish whether the volunteer students were representative of the general student population in terms of mental wellbeing. Students who perceive themselves as being stressed, anxious or depressed may have been more likely to volunteer to participate. There was no mental health screening of students before participation. As the students are on clinical courses all have occupational health clearance to exclude active major mental illnesses (such as psychosis). Experienced staff were available to supervise the project with clear pathways to access support available.

The students who volunteered were from the full range of health-care courses. There was no control arm, so a placebo effect cannot be ruled out for all measures. There was no economic analysis or environmental analysis in terms of sustainability. We acknowledge that blood pressure and pulse rate are poor physiological proxy measures of stress; more definitive measures such as blood or salivary cortisol levels are contentious and beyond the scope of this study.

The main limitation of the study was the lack of opportunity for a two week or one month follow-up because of student term dates. The results of the GAD7 and other scales should therefore be viewed as indicators of potential benefit to explore in future. The daily analogue scales were promising in terms of short-term benefit, but the limitations of a lack of a control arm apply. A further limitation was that cybersickness was assessed by simply asking the students rather than using a validated scale. Any other adverse effects were monitored by asking simply by asking students whether they had had any problems or issues. In the same way, practical issues such as the space used, and any technical problems experienced were assessed only through the qualitative feedback from the students rather than any more formal evaluation.

The qualitative data was elicited through responses to free-text open questions, which is limited in its richness to a description of the students’ experience. However, its findings will inform the next stages in this programme of research.

Future research

The positive response by students to this pilot experience suggests that a further trial with a control arm should be explored. Prior to the feasibility trial, no prespecified criteria were set regarding whether to proceed with a definitive trial as potential problems were to some extent unknown. The student feedback suggests that a further trial is indicated, with a larger sample, a control arm and with a more extensive follow-up of students to establish potential lasting effects of the intervention.

At a time when there is so much concern about student mental health, any interventions that may help with stress reduction need further evaluation. Interventions could be targeted at times that may be more stressful such as exam time, or when clinical students commence their placements on the wards. There is a need for studies with larger samples, a control group and with more in-depth qualitative analysis.

Implications for practice

It may be that universities wish to provide similar applications in the future. As VR is increasingly being used in teaching, familiarity with it and infrastructure to support it should increase. There is an increasing awareness that staff as well as students at university are under greater stress than the general population (Morrish, 2019). As such, it would seem reasonable to offer such interventions to staff as well.

Conclusions

The students in this feasibility study were very positive about this VR hypnotherapy experience. There were improvements on measures of wellbeing over the three days of the intervention. These encouraging results suggest that a larger study with a control arm is warranted. At a time when student welfare is a major concern, this represents a promising addition to the offer universities can provide.

Figures

Figure 1.

Mean Likert scale scores gauging participants’ present levels of stress (1a), happiness and sadness (1b) and calmness and anxiety (1c) pre- and post-VR hypnosis for each session on Figures 1(a), 1(b) and 1(c)

Figure 2.

Mean participant blood pressure (2a) and pulse rate per minute (2b) at pre- and post-VR hypnosis for each session

Table 1.

Pre- and post-intervention scores for perceived stress, wellbeing and mood measures and pre- and post-session values (averaged across sessions) for Likert scales, blood pressure and pulse rate measures with difference in means and Hedges’ g effect sizes

	Pre-intervention	Post-intervention	Pre- versus post-intervention
	Mean (SD)	Mean (SD)	Difference in means (95% CI)	p	Hedges’ g
PSS (0−40)	20.27 (3.92)	19.87 (2.88)	0.40 (−1.25, 2.05)	0.610	0.13
WEMWBS (14–70)	43.47 (9.14)	45.87 (8.89)	2.40 (1.33, 3.53)	0.006	0.76
PHQ-2 (0–6)	2.00 (1.73)	1.27 (1.49)	0.73 (0.40, 1.07)	0.010	0.83
GAD-7 (0–21)	8.53 (4.81)	7.13 (4.69)	1.40 (−0.18, 1.90)	0.078	0.47
	Pre-session	Post-session	Pre- versus post-session
Likert scale (0–10)	Mean (SD)	Mean (SD)	Difference in means (95% CI)	p	Hedges’ g
Stress	4.78 (1.35)	2.82 (1.33)	1.96 (1.43, 2.48)	<0.001	1.96
Happy	5.47 (1.70)	6.20 (1.99)	0.73 (0.36, 1.11)	0.008	0.94
Sad	3.16 (2.67)	2.66 (2.22)	0.50 (0.18, 0.82)	0.005	0.82
Calm	5.69 (1.14)	7.56 (1.49)	1.87 (1.13, 2.61)	<0.001	1.32
Anxiety	4.73 (1.59)	2.87 (1.53)	1.87 (1.35, 2.39)	<0.001	1.88
Blood pressure
Systolic	114.47 (13.61)	112.20 (14.82)	2.27 (−1.19, 5.72)	0.181	0.34
Diastolic	73.62 (10.01)	74.13 (11.01)	−0.51 (−2.78, 2.20)	0.703	−0.53
Pulse (rate/minute)	76.62 (8.98)	74.04 (7.34)	2.58 (0.84, 4.35)	0.021	0.56

Notes:

CI = confidence interval; SD = standard deviation; WEMWBS = Warwick-Edinburgh mental wellbeing scale; PSS = Perceived stress scale; PHQ-2 = Patient health questionnaire-2; GAD-7 = Generalized anxiety disorder-7 {Where variables did not meet requirements for univariate normality using skewness and kurtosis estimates [WEMWBS, PHQ-2, Happy (Likert scale), Diastolic blood pressure and Pulse rate], bias corrected and accelerated (Bca) bootstrapping using 2,000 replications was used}; Hedges’ g: small effect size = 0.2; medium = 0.5; large = 0.8 (26); positive difference in means and effect size represent an improvement on the relevant measure. Significant differences are highlighted in italic

Source: Authors’ own creation