The use of virtual reality to improve quality of recovery in women undergoing gynecological surgeries: a randomized controlled trial

Abstract

Virtual reality (VR) is an emerging non-pharmacological intervention that has shown promise in reducing anxiety and pain, but its effect on post-operative recovery is less studied. We assessed the use of VR on postoperative recovery, anxiety, and pain in women undergoing gynecological surgery. This was a randomized controlled trial that recruited women undergoing gynecological surgery. Patients were randomized into two groups: no intervention; or VR calming scenarios with meditation and breathing exercises administered perioperatively. The primary outcome would measure the quality of recovery through the Quality of Recovery-40 (QoR-40) total scores at postoperative 24 h. Secondary outcomes included the perioperative visual analog scale-anxiety (VAS-A) scores, pain scores, and analgesic use. A total of 104 participants were randomized equally into the VR and control groups with no dropout. The VR group showed significantly greater QoR-40 total scores at postoperative 24 h (mean difference (MD) 7.23, 95% CI 1.44–13.02, p = 0.015), with notable improvements in physical comfort (MD 2.58, p = 0.015), and pain (MD 1.38, p = 0.005) dimensions when compared with control group. Anxiety scores were significantly lower in the VR group during preoperative period (MD 0.96, p = 0.034), and at postoperative 48 (MD 0.89, p = 0.01) and 72 h (MD 0.66, p = 0.03). Patient satisfaction with VR was high, with > 40% of patients reporting satisfaction of “good” or “excellent”. VR significantly improved postoperative recovery scores at 24 h and was associated with reduced anxiety, pain, and analgesic use. While the improvements were statistically significant, the clinical relevance of the recovery score difference remains uncertain. These findings support the potential role of VR as a non-pharmacological adjunct to enhance recovery in gynecological surgery.

Trial registration: This study was registered on clinicaltrials.gov registry (NCT03685422) with date of first registration on 24/09/2018.

Introduction

Postoperative recovery plays a vital role in surgical outcomes, influencing both patient well-being and healthcare resource utilization. Effective recovery management encompasses not only physical aspects but also psychological and social dimensions¹. While clinicians typically focus on recovery metrics such as the absence of complications or length of hospital stay, patients often define recovery as the ability to return to daily life without pain². Given these differing perspectives, precise tools to measure recovery outcomes are critical for both clinical care and research.

The Quality of Recovery-40 (QoR-40) has been extensively validated for assessing recovery across various surgical settings and provides a comprehensive measure capturing multiple dimensions of recovery³. Developed to assess early postoperative recovery, it has demonstrated high reliability and validity across diverse clinical contexts. A systematic review of 17 studies, encompassing over 3400 patients, confirmed the QoR-40’s strong internal consistency (pooled Cronbach Alpha = 0.91) and excellent test-retest reliability and inter-rater reliability, supporting its consistency across settings⁴. Another systematic review identified the QoR-40 as the only tool meeting eight essential criteria, including validity, reliability, and feasibility, making it highly suitable for ambulatory surgical environments⁵. Moreover, it has shown particular applicability in gynecological surgery, validated for procedures like hysteroscopy⁶, with findings suggesting age-related variations in postoperative recovery, notably with younger patients reporting lower pain scores⁷. However, despite the QoR-40’s robust applicability, managing postoperative pain remains challenging, especially in major gynecological surgeries where complications such as nausea, infections, and prolonged pain are common⁸.

As surgical care evolves, non-pharmacological approaches like virtual reality (VR) are gaining traction as tools to enhance patient experiences⁹. VR has shown promise in managing pain, easing anxiety, and improving outcomes across various medical settings, including preoperative care¹⁰. We previously conducted a preliminary study exploring the feasibility of VR for reducing preoperative anxiety among gynecological patients, which demonstrated significant improvements and positive patient feedback¹¹. Beyond preoperative use, VR therapy has also been reported to alleviate persistent pain, accelerate wound healing, and enhance neurorehabilitation in patients with burns and complex regional pain syndrome^12,13. The technology typically comprises auditory components (such as earphones or headphones), visual displays (head-mounted units), and integrated setups like motion tracking systems. By engaging multiple sensory inputs, VR offers an immersive experience that fosters a strong sense of presence in the virtual environment^14–16. However, its impact on postoperative recovery, particularly as measured by the QoR-40, remains understudied.

In this study, we hypothesized that VR administered before and after surgery would improve the quality of recovery at postoperative 24 h in women undergoing gynecological surgery. We also assessed the impact of VR intervention on the quality of recovery, pain scores, and anxiety before surgery and beyond postoperative 24 h.

Methods

We previously conducted a preliminary cohort study to assess the feasibility of using VR in our institution¹¹. To further evaluate the effectiveness of the VR intervention compared to routine clinical practice, we conducted a randomized controlled trial on women undergoing gynecological surgery at KK Women’s and Children’s Hospital, Singapore between August 2022 and February 2024. This study was approved by the SingHealth Centralized Institutional Review Board (reference number CIRB 2018–2200; approved on 3 May 2018). Both the first (preliminary cohort study) and second (randomized controlled trial) phases were registered on Clinicaltrials.gov (NCT03685422) with date of first registration on 24/09/2018¹¹. The study methodologies for the second phase were developed in accordance with the Consolidated Standards of Reporting Trials (CONSORT) standards¹⁷.

Patients were recruited during their consultations in pre-anesthetic clinics, during ward stays prior to surgery, or in pre-admission areas. Study investigators identified potential participants by reviewing surgical schedules and assessing eligibility through patients’ medical records. Women aged 21–70 years, classified as American Society of Anesthesiologists (ASA) physical status I or II, with no visual impairments and scheduled for gynecological surgery, were considered eligible. Exclusion criteria included severe motion sickness, significant respiratory disease, obstructive sleep apnea, current pregnancy, or inability to communicate in English or comprehend the study questionnaires. Eligible patients were approached in the preoperative setting, where study procedures, risks, and benefits were explained, and written informed consent was obtained.

Randomization was conducted through sealed opaque envelopes generated by computerized random number tables. Before surgery, patients completed a baseline assessment that included the QoR-40 and EuroQol Five-Dimensional Three-Level (EQ-5D-3L) questionnaires, as well as demographic information such as age and race^3,18. Pain scores were measured separately using a Numerical Rating Scale (NRS). Baseline anxiety was using the State-Trait Anxiety Inventory (STAI)¹⁹, and a visual analog scale-anxiety (VAS-A) on a 0–100 mm rating scale (0 being ‘not anxious at all’; 100 being ‘extremely anxious’)²⁰.

In the VR group, the assigned patients underwent VR intervention using the Oculus Quest 2 headset (Reality Labs, California, USA) and chose from nine calming scenarios featuring meditation music and breathing exercises within the Nature Treks VR (Greener Games, Telford, Shropshire, Great Britain) application on the MetaQuest 2 VR platform (Meta, Menlo Park, California, United States of America). Each VR intervention lasted 10–20 min in a quiet pre-operative area, followed by reassessment of anxiety, pain scores. The choice of scenario, satisfaction scores on the VR experience and VAS-A scores were measured after the VR exposure preoperatively. No intervention was administered in the control group (routine clinical practice); instead, they completed only the standard assessments (anxiety and pain scores) 30 min after the baseline survey.

During the intraoperative period, data was collected on preoperative medications (if any), anesthesia start time, duration of surgery and intraoperative analgesic use. The intraoperative and postoperative care proceeded as per hospital standards at the discretion of the clinical team. In the post-anesthesia care unit (PACU), data was collected on the duration of stay, postoperative analgesic use, minimum and maximum pain scores at rest. No VR intervention was administered in PACU.

During postoperative 0–24 h, patients in the VR group were asked to engage in three VR interventions, with each session lasting between 10 and 20 min. Any adverse events, such as nausea or vomiting, were monitored; and the VR intervention was discontinued if any severe symptoms occurred. Follow-up assessments were performed in both groups including QoR-40 scores, anxiety, pain scores, analgesic use, and any side effects. The assessments were conducted at postoperative fixed time points of 24, 48-, and 72 h after each VR exposure, to minimize participant burden and avoid disrupting routine care. For those discharged earlier, follow-ups were conducted via an online survey or phone call.

Statistical analysis

A sample of 104 (52 × 2) was required to achieve 80% power to reject the null hypothesis of equal means when the population mean difference in QoR-40 total score at postoperative 24 h was 15.0 with a same standard deviation (SD) of 25.3 with significance level (alpha) of 5%, 1:1 allocation ratio, withdrawal rate of 10% and using a two-sided two-sample equal-variance t-test. The calculation was based upon Hong et al. on mean QoR-40 total score in gynecological patients of 143 (SD 25.3), and studies by Myles et al. that reported a difference in QoR-40 total score of ≥ 10 points being considered as clinically significant and equivalent to a 15% improvement in recovery^7,21,22. Hence, we chose a 15-point difference in QoR-40 as clinically relevant, consistent with our previous feasibility study¹¹.

All demographic and clinical variables were summarized based on the VR and control groups. Categorical and continuous variables were summarized as frequency (percentages) and mean (SD) or median (interquartile range (IQR)), whichever appropriate, respectively. The primary outcome QoR-40 total score at postoperative 24 h was treated as a continuous variable, and difference in QoR-40 total score between VR and control groups were tested using two sample independent t-test and were reported as mean difference (MD) with 95% confidence interval (CI).

The secondary outcomes which included scores in other QoR-40 dimensions, VAS-A score, pain scores (at rest and with movement) at all other time points, were treated as continuous data and were summarized as mean (SD). Difference in outcomes between VR and control groups were tested using two sample independent t-test and were reported as mean difference (MD) with 95% confidence interval (CI). Differences between VR and control groups at each time point were estimated using regression based on repeated measures mixed models (repeated ANOVA), which accounts for the dependence among repeated measurements on the same patient. Mixed-models for continuous data was employed, in SAS® version 9.4 (SAS Institute, Cary, North Carolina; PROC MIXED), with time, group and time by group interaction as fixed-effects, and time as a random-effect with an unstructured (general) variance-covariance matrix. The estimation method was based on a residual (restricted) maximum likelihood technique and the variance-covariance matrix of the parameter estimates computed using a sandwich (empirical) estimator. All tests were two-sided, unless otherwise stated. Significance level was set at 5%. All analyses were performed using SAS® version 9.4 (Cary, North Carolina: USA).

Results

A total of 104 participants were enrolled and randomized equally into the VR intervention or control (no VR intervention) group (Fig. 1). No severe VR-related adverse events were observed; all VR participants competed pre- and postoperative VR sessions, and no sessions were discontinued due to symptoms. No participants withdrew from the study, and all completed the postoperative follow-up. There were no significant differences in baseline characteristics between the two groups except in the QoR-40 dimensions of physical independence and physical support (Tables 1 and 2). Physical independence scored higher in the experimental group pre-intervention (MD 0.33, 95% CI 0.02–0.65, p = 0.039), while physical support scored higher in the control group pre-intervention (MD 0.52, 95% CI 0.05–0.99, p = 0.030).

Fig. 1.

CONSORT flow diagram. CONSORT consolidated standards of reporting trials.

Table 1.

Patient demographic characteristics.

Variable	Experimental (n = 52)	Control (n = 52)
Age (years), mean (SD)	35.5 (7.1)	35.8 (8.0)
Race, n (%)
Chinese	35 (67.3)	35 (67.3)
Malay	12 (23.1)	12 (23.1)
Indian	1 (1.9)	4 (7.7)
Others	4 (7.7)	1 (1.9)
ASA status, n (%)
I	49 (94.2)	43 (82.7)
II	3 (5.8)	9 (17.3)
Weight (kg), mean (SD)	62.3 (11.8)	59.9 (12.9)
Height (m), mean (SD)	1.6 (0.1)	1.6 (0.1)
BMI (kg m− 2), mean (SD)	24.6 (4.3)	23.8 (5.0)
Nature of surgery, n (%)
Cystectomy	25 (48.1)	21 (40.4)
Myomectomy	23 (44.2)	23 (44.2)
Dilatation and curettage	5 (9.6)	1 (1.9)
Hysterectomy	2 (3.9)	3 (5.8)
Others	3 (5.8)	7 (13.5)
Surgical approach, n (%)
Vaginal	11 (21.2)	10 (19.2)
Open	19 (36.5)	22 (42.3)
Minimally invasive (laparoscopic)	30 (57.7)	28 (53.8)
Preoperative
Medication, n (%)	13 (25.0)	9 (17.31)
Gemeprost	8 (15.4)	5 (9.6)
Misoprostol	5 (9.6)	4 (7.7)

ASA American society of anesthesiologists; BMI body mass index; SD standard deviation; VR virtual reality.

Table 2.

Patient baseline anxiety data.

Variable	Experimental (n = 52)	Control (n = 52)
STAI state anxiety (20–80), mean (SD)	35.8 (9.0)	37.3 (9.4)
STAI trait anxiety (20–80), mean (SD)	33.2 (8.9)	32.1 (9.1)
STAI total score (40–160), mean (SD)	69.1 (15.3)	69.4 (16.5)
EQ-5D-3 L VAS (0-100), mean (SD)	74.7 (24.4)	68.2 (28.5)

EQ-5D-3 L EuroQol five-dimensional three-level; SD standard deviation; STAI State-trait anxiety inventory; VAS-A visual analog scale-anxiety.

QoR-40 scores across all time points are illustrated in Fig. 2. The primary outcome of QoR-40 total scores at postoperative 24 h was significantly greater in the experimental group following the third VR intervention (MD 7.23, 95% CI 1.44–13.02, p = 0.015; Fig. 2a). Adherence to the postoperative VR protocol was complete, as all VR-arm participants completed the three sessions within 24 h.

Fig. 2.

The longitudinal analyses for (a) the primary outcome of QoR-40 total score; and QoR-40 dimensions of (b) physical comfort; (c) physical independence; (d) pain; (e) emotional status; (f) physical support; in VR and non-VR groups. QoR-40 quality of recovery-40.

For secondary outcomes, notable improvements were observed in the QoR-40 dimensions of physical comfort (MD 2.58, 95% CI 0.50–4.65, p = 0.015), and pain (MD 1.38, 95% CI 0.42–2.34, p = 0.005) following the third VR intervention. There is a general drop in scores from the baseline to the scores at postoperative 0–24 h and a progressive rise in scores after. In addition, scores in QoR-40 dimensions of physical comfort (Fig. 2b), physical independence (Fig. 2c), and pain (Fig. 2d) were higher in the experimental group compared to the control group. Scores in QoR-40 dimension of emotional status were lower in the experimental group at baseline and at first intervention postoperative 0–24 h but progressive increased and remained higher after the second intervention within 24 h (Fig. 2e). Scores in QoR-40 dimension of physical support were generally lower in the experimental group except after the second and third intervention within 24 h (Fig. 2f).

As seen in Fig. 3, Preoperative VAS-A anxiety scores were significantly lower in the experimental group compared to the control group (MD 0.96, 95% CI 0.07–1.85, p = 0.034), and at postoperative 48 (MD 0.89, 95% CI 0.21–1.56, p = 0.01) and postoperative 72 h (MD 0.66, 95% CI 0.07–1.26, p = 0.03). VAS-A scores across all time points are illustrated in Fig. 3a. VAS-A scores in both groups were lower postoperatively compared to preoperatively. The VAS-A scores started higher in the experimental group compared to the control group but subsequently was consistently lower in the experimental group. Immediate pre-/ post-session anxiety was obtained only for the preoperative VR exposure. Pain scores (at rest and on movement) across all time points showed a general spike in scores in both groups at postoperative 24 h followed by a gradual decrease thereafter (Fig. 3b and c) The difference in pain scores between the two groups were not significant. The dosage of paracetamol used was also lower in the experimental group within postoperative 24 h (p < 0.001), as shown in Table 3. Mixed‑effects analyses revealed significant group × time interactions for all QoR‑40 dimensions (Fig. 2), as well as for anxiety and pain outcomes (Fig. 3) (all p < 0.001), reflecting significantly different temporal response patterns between the two groups.

Fig. 3.

The longitudinal analyses for the secondary outcomes of (a) VAS-A; (b) pain scores (at rest); and (c) pain scores (at movement) in VR and non-VR groups. VAS-A visual analog scale—anxiety.

Table 3.

Patient pain and psychological characteristics.

Variable	Experimental (n = 52)	Control (n = 52)	p
Intraoperative
Analgesics, n (%)
Morphine	50 (96.2)	48 (92.3)	0.400
If morphine was used, the total dosage (mg), mean (SD)	6.7 (2.1)	110.6 (721)	0.323
Fentanyl	49 (94.2)	48 (92.3)	0.696
If fentanyl was used, the total dosage (mcg), mean (SD)	101.5 (20.0)	117.7 (131)	0.400
Paracetamol	42 (80.8)	39 (75.0)	0.478
If paracetamol was used, the total dosage (mg), mean (SD)	1000 (0.0)	994.9 (32.0)	0.324
Others	8 (15.4)	10 (19.2)	0.604
Duration of surgery (minutes), mean (SD)	119.6 (54.8)	126.6 (60.9)	0.538
Postoperative in PACU
Duration of PACU stay (mins), mean (SD)	85.8 (30.8)	90.0 (32.7)	0.501
Minimum pain score in PACU (at rest, 0–10), mean (SD)	1.3 (1.8)	1.2 (1.9)	0.751
Maximum pain score in PACU (at rest, 0–10), mean (SD)	4.0 (3.5)	3.4 (3.1)	0.422
Analgesics used in PACU, n (%)
Morphine	13 (25.0)	13 (25.0)	1.000
If morphine was used, the total dosage (mg), mean (SD)	4.4 (2.0)	3.9 (2.1)	0.602
Fentanyl	5 (9.6)	2 (3.9)	0.240
Paracetamol	4 (7.7)	1 (1.9)	0.169
Others	3 (5.8)	3 (5.8)	1.000
Postoperative 0–24 h (after 1st VR experience)
Analgesics, n (%)
Morphine	2 (3.8)	2 (3.8)	1.000
If morphine was used, the total dosage (mg), mean (SD)	3.5 (0.7)	27.4 (11.5)	0.209
Fentanyl	0 (0.0)	0 (0.0)
Paracetamol	46 (88.5)	46 (88.5)	1.000
If paracetamol was used, the total dosage (mg), mean (SD)	1233.7 (436.0)	1771.7 (664.2)	< 0.001
Others	17 (32.7)	18 (34.6)	0.836
Postoperative 0–24 h (after 2nd VR experience)
Analgesics, n (%)
Morphine	0	0	–
Fentanyl	0	0	–
Paracetamol	19 (36.5)	0	< 0.001
Others	3 (5.8)	0	0.079
Postoperative 0–24 h (after 3rd VR experience)
Analgesics, n (%)
Morphine	1 (1.9)	0	0.315
Fentanyl	0	0	–
Paracetamol	21 (40.4)	0	< 0.001
Others	4 (7.7)	0	0.041
Postoperative 25–48 h
Analgesics, n (%)
Morphine	0 (0.0)	0 (0.0)	–
Fentanyl	0 (0.0)	0 (0.0)	–
Paracetamol	45 (86.5)	46 (88.5)	0.767
If paracetamol was used, the total dosage (mg), mean (SD)	1927.8 (758.7)	1858.7 (735.3)	0.660
Others	11 (21.2)	13 (25.0)	0.642
Postoperative 49–72 h
Analgesics, n (%)
Morphine	0 (0.0)	0 (0.0)	–
Fentanyl	0 (0.0)	0 (0.0)	–
Paracetamol	42 (80.8)	38 (73.1)	0.352
If paracetamol was used, the total dosage (mg), mean (SD)	1904.8 (842.8)	1789.5 (851.3)	0.352
Others	8 (15.4)	3 (5.8)	0.111

PACU post anesthesia care unit; QoR-40 quality of recovery-40; SD standard deviation; STAI state-trait anxiety inventory; VAS visual analog scale; VAS-A visual analog scale- anxiety; VR virtual reality.

In the preoperative period, 21 (40.4%) patients rated their experience as ‘good’ and 27 (51.9%) rated as ‘excellent’ (Table 4). After the first post-operative VR, 22 (42.3%) patients rated ‘good and 27 (51.9%) ‘excellent’. After the second post-operative VR, 24 (46.2%) patients rated ‘good’ and 25 (48.1%) ‘excellent’. After the third post-operative VR, 24 (46.2%) patients rated ‘good’ and 24 (46.2%) rated ‘excellent’. The ‘blue deep/blue moon/blue ocean’ scenarios were the most popular (Table 4).

Table 4.

VR experience in VR group.

Variable	Preoperative	Postoperative 1st VR experience	Postoperative 2nd VR experience	Postoperative 3rd VR experience
Scenarios, n (%)a
Black beginning	5 (9.6)	7 (13.5)	3 (5.8)	6 (11.5)
Blue deep/ blue moon/ blue ocean	34 (65.4)	16 (30.8)	13 (25.0)	17 (32.7)
Bamboo green/ meadow green/ Jade Jurassic	12 (23.1)	13 (25.0)	17 (32.7)	20 (38.5)
White winter	9 (17.3)	8 (15.4)	13 (25.0)	7 (13.5)
Red fall/ red savanna/ orange sunset	10 (19.2)	10 (19.2)	12 (23.1)	4 (7.7)
Satisfaction with VR, n (%)b
Poor	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Fair	4 (7.7)	3 (5.8)	3 (5.8)	3 (5.8)
Good	21 (40.4)	22 (42.3)	24 (46.2)	24 (47.1)
Excellent	27 (51.9)	27 (51.9)	25 (48.1)	24 (47.1)

QoR-40 quality of recovery-40; SD standard deviation; VAS-A visual analog scale- anxiety; VR virtual reality.

^aPatients may choose more than one scenario for each VR experience.

^bMissing one patient data on satisfaction for postoperative 3rd VR experience.

Discussion

This randomized controlled trial demonstrated the potential of virtual reality (VR) in enhancing postoperative recovery in women undergoing gynecological surgery. Statistically significant improvements in quality of recovery scores were observed in the VR group compared to the control group receiving routine clinical care at 24 h postoperatively, particularly in the domains of physical comfort and pain. However, as the confidence interval for the mean difference included values below the commonly accepted minimal clinically important difference, the clinical relevance of this improvement should be interpreted with caution. Although the VR group had higher baseline anxiety scores, they reported significantly lower anxiety levels than the control group at each postoperative time point, from the immediate postoperative period through 72 h. High patient satisfaction with the VR experience also indicates its acceptability and potential as an effective adjunct to postoperative care.

The QoR-40 measurements across perioperative time points suggest that VR may support better recovery, with statistically significant improvements particularly in the domains of physical comfort and pain. The primary outcome—the improvement in QoR-40 total scores at 24 h postoperatively, notably following the third VR intervention—demonstrates a potential cumulative benefit. These findings align with previous studies^5–7, suggesting that the benefits during recovery extend beyond physical symptom relief to support overall recovery quality. Tashjian et al. found that VR use significantly reduced discomfort in hospitalized patients, aligning with our findings in the postoperative setting²³. Similarly, Li et al. emphasized the broad utility of VR in pain management, showing its potential to ease discomfort by diverting attention and reducing stress⁹.

For pain management, the use of VR as a distraction tool significantly reduced perceived pain intensity during medical procedures. Hoffman et al. demonstrated this in burn patients, where analgesic effects of VR remained consistent across multiple sessions²⁴. Malloy and Milling likewise reported that VR distraction reduced pain across various procedural contexts, supporting VR as a promising non-pharmacological option for postoperative care²⁵. In our study, improvements became more pronounced after the third VR exposure, suggesting a cumulative effect. Hoffman et al. also reported sustained analgesic effects with repeated use, indicating that ongoing VR exposure may be necessary to maximize recovery benefits²⁴.

The pain-modulating effects of VR remain a topic of active research due to the complex nature of pain as shaped by both psychological and neural factors, making it difficult to address with non-pharmacological treatments. According to Melzack’s neuromatrix theory of pain, pain arises not only from sensory input but from a complex “body-self neuromatrix” that generates characteristic “neurosignatures”, triggered by injury, pathology, or stress²⁶. This underscores pain as a multidimensional experience, and suggest that VR may modulate pain by influencing this neural network. A systematic review by Baker et al. covering 70 studies with over 4100 participants in acute and chronic pain found that VR significantly improved pain intensity and quality, especially in acute pain settings²⁷. Similarly, Hoffman et al. observed through functional magnetic resonance imaging (fMRI) that VR could reduce activity in pain-related brain regions, such as the insula and thalamus²⁸. In their study combining VR and opioids, VR alone reduced pain and unpleasantness, with additional benefit when paired with opioids, supporting VR as an adjunct to pharmacological pain relief²⁸.

The impact of VR on pain scores in our trial was modest, with no significant differences in pain at rest or during movement between the experimental and control groups. However, patients in the VR group required fewer analgesics within the first 24 h post-surgery, suggesting a potential role of VR in improving pain coping mechanisms. This aligns with studies by Wiederhold et al.²⁹ and Deo et al.³⁰ which demonstrated that VR could reduce reported pain and anxiety levels during gynecological procedures and outpatient hysteroscopy. These findings suggest VR can support pain management indirectly by enhancing coping strategies and reducing stress.

Notably, the pain dimension of the QoR-40 survey showed significant improvement in the experimental group, despite the lack of difference in direct pain scores. This likely reflects the multidimensional nature of the QoR-40 pain domain, which encompasses not only pain intensity but also its subjective impact on recovery, including discomfort, distress, and coping ability. Unlike direct pain scores that capture intensity at specific moments, the QoR-40 pain domain reflects broader psychosocial and emotional factors that VR may positively influence.

The immersive and distracting effects of VR may reduce the perceived burden of pain and promote relaxation, improving overall well-being even if the physiological experience of pain intensity remains unchanged. This aligns with evidence that VR can enhance pain tolerance and alters perception by mitigating stress and fostering a positive emotional state³⁰. These findings highlight the need to examine how VR interventions can improve recovery holistically, beyond direct measure of pain intensity.

Although emotional status scores in the experimental group were initially lower, they progressively improved after the first VR intervention, as reflected in QoR-40 scores at 24 h postoperatively. In a study by Chirico et al.³¹ VR was shown to alleviate psychological distress and anxiety in cancer patients, illustrating its capacity to enhance emotional well-being even in challenging health contexts. Although this study focused on cancer patients, these findings suggest that VR’s ability to relieve distress may also extend to other recovery phases, such as the postoperative period. Supporting this, a review by Donnelly et al. explored the effectiveness of VR exposure therapy (VRET) for anxiety disorders across diverse settings and found VR to be a valuable tool in reducing anxiety, even for conditions like Post traumatic stress disorder (PTSD) and social anxiety³². In our trial, anxiety levels, measured by VAS-A, were most significantly reduced at 48–72 h postoperatively.

During the VR intervention, participants could select their VR scenarios, with the blue-themed scenarios being the most popular, followed by the green-themed. Blue and green have been associated with reduced anxiety on STAI scores, whereas red is associated with higher STAI scores, in an experimental study of 40 undergraduate students examining the effects of the four primary colors on anxiety^33,34. A study by White et al.³⁵ which explored the effect of green and blue spaces on mental health in an adjusted sample of 16,307 participants across 18 countries, found a positive association between the frequency of visits to green spaces in the past four weeks and positive well-being, and a negative association with mental distress and use of prescribed depression medication (green spaces included areas classified as forests, grassland, shrubland and cultivated land). After controlling for frequency of green space visits, visits to inland-blue and costal-blue spaces were also positively associated with positive well-being and lower rates of mental distress (inland-blue space: areas classified as water bodies, wetlands; coastal-blue spaces included beaches, rocky shores etc.)³⁵. With regards to anxiety, increased neighborhood green space was significantly associated with lower levels of anxiety^36,37.

We postulate that VR scenarios with different color themes may have varying effects on psychological factors like anxiety, which could have downstream effects on quality of recovery and pain. This could help explain the popularity of blue and green-themed scenarios, given their psychological benefits. However, the effects of the different individual VR scenarios have yet to be studied. Future work should examine how specific VR scenarios may impact these outcomes, including participants’ perceptions of VR, and how we may tailor these experiences to better suit the local context.

The strength of this study lies in its randomized controlled design and the use of a validated outcome measure in the QoR-40. Protocol adherence was high, with universal completion of three postoperative VR sessions within 24 h, supporting feasibility in routine perioperative workflows. This strengthens internal validity and enhances the robustness of our findings. However, several limitations warrant consideration. Firstly, baseline differences were noted in the QoR-40 dimensions of physical independence and physical support. These may have arisen due to chance or unmeasured factors such as pre-existing resilience, physical activity levels, or available support systems. While these small baseline variations may have influenced early outcomes, their effects are likely mitigated across repeated measurements, allowing a more accurate estimation of VR’s true impact on recovery. Secondly, expectation bias may have contributed to improved perceived recovery, as patients aware of receiving VR could have experienced a psychological uplift³⁸. While this is an inherent limitation of non-blinded interventions, it reflects real-world conditions in which the perceived value of treatment can influence outcomes. Thirdly, the study population comprised predominantly Chinese and Asian women undergoing gynecological surgery, which may limit the generalizability of findings to other ethnicities, genders, or surgical contexts.

In terms of anxiety measurement, the use of the VAS-A and State-Trait Anxiety Inventory (STAI) provided a useful snapshot; however, the full version of STAI was administered preoperatively to minimize respondent burden in the immediate postoperative period. Consequently, postoperative anxiety trends were evaluated using VAS-A at multiple time points, which limits direct comparability with STAI and precludes analysis of postoperative STAI trajectories. Immediate pre/post anxiety was not obtained around each postoperative session, precluding analysis of within-session postoperative changes. Future studies should include repeated postoperative STAI to triangulate anxiety trajectories with brief, and comprehensive measures and incorporate brief pre/post postoperative measures to characterize session-level effects. External factors such as variability in clinical care, hospital environment, or interpersonal interactions were also not controlled for, and may have influenced anxiety or recovery outcomes. We also did not collect postoperative length of stay (LOS), which limits our ability to relate early recovery improvements to discharge outcomes and resource utilization. Given potential non-clinical influences on LOS (e.g. bed availability, weekend discharges, social support), future studies should include LOS and time-to-readiness for discharge as predefined endpoints to strengthen the link between patient-centered recovery and system-level outcomes.

Cultural relevance is an important consideration. Although the VR intervention was designed for relaxation, its content may not have been equally relatable to all patients. Using localized or culturally tailored VR content in future studies could improve patient engagement and therapeutic benefit. The study also did not examine alternative types of VR applications or settings, which may affect the generalizability of the findings to other modalities. To assess real-world effectiveness, we used a control group without VR, recognizing the potential for expectation bias but avoiding the ethical and practical challenges of sham VR. In clinical practice, patients either experience VR or they do not. There is no “inert” VR experience. Allowing participants to choose their VR content reflects practical use and increases ecological validity.

Within the first 24 h postoperatively, participants in the VR group completed three sessions with immediate anxiety and pain assessments, whereas the control group underwent a single assessment. This asymmetry in assessment frequency and timing may introduce measurement-related bias, including proximity to the intervention, variation in analgesic administration, and disrupted clinical workflow. Increased interaction with staff and repeated survey completion in the VR group may also have led to attention or assessment reactivity, potentially influencing outcomes independently of the VR intervention. Future trials should harmonize assessment schedules (e.g. fixed-time sampling and parallel assessment frequency) across group to minimize these biases and to permit more robust dose-response analyses.

In addition, detailed VR usage metrics and content-outcome relationships were not collected, precluding evaluation of whether participant preferences or specific VR scenarios moderated the VR effect. We also did not implement a dedicated VR-specific adverse event log. Although no severe adverse events were observed, minor symptoms were not prospectively captured or categorized. Future studies should incorporate structured monitoring of VR-related adverse events, including symptom type, severity, timing to allow more comprehensive evaluation of safety. Finally, the exclusion of VR exposure from the control group minimized potential confounding arising from passive distraction or the novelty of headset use. This approach is supported by a recent systematic review indicating that passive control conditions do not systematically inflate VR effect sizes, while active control interventions do not consistently outperform routine care³⁹. Collectively, these findings support the validity of the chosen control design.

Conclusion

In conclusion, VR shows promise as a non-pharmacological adjunct to improve quality of recovery in the immediate postoperative period. This trial highlights the potential of VR in fostering a more holistic recovery process, improving both physical and emotional dimensions of patient care. As an innovative tool in perioperative management, VR may contribute to a more patient-centered recovery approach, enhancing overall satisfaction and well-being.

Abbreviations

ASA

American society of anesthesiologists

CI

Confidence interval

CONSORT

Consolidated standards of reporting trials

EQ-5D-3L

EuroQol five-dimensional three-level

fMRI

Functional magnetic resonance imaging

IQR

Interquartile range

LOS

Length of stay

MD

Mean difference

NRS

Numerical rating scale

PACU

Post-anesthesia care unit

QoR-40

Quality of recovery-40

SD

Standard deviation

STAI

State-trait anxiety inventory

VAS-A

Visual analog scale-anxiety

VR

Virtual reality

Author contributions

JJIC contributed to conceptualization, methodology, formal analysis, funding acquisition, writing of the original draft, manuscript review, and editing. RS contributed to conceptualization, methodology, formal analysis, software, manuscript review, and editing. YTRH contributed to conceptualization, methodology, formal analysis, funding acquisition, manuscript review, and editing. MHSAM and CWT contributed to conceptualization, methodology, investigation, formal analysis, manuscript review, and editing. BLS contributed to conceptualization, methodology, investigation, formal analysis, funding acquisition, manuscript review and editing. All authors read and approved the final manuscript.

Funding

JJIC is supported by SingHealth Duke-NUS Academic Medicine (AM) Research Grant (Reference number AM/SU065/2022). YTRH and BLS are awarded the Academic Medicine—Enhancing Training, Healthcare, Outcomes and Standards (AM-ETHOS) Duke-NUS Medical Student Research Fellowship Award year 2024-25 (ref no. AM-ETHOS01/FY2024/17-A17). The funders had no role in the conceptualization, design, data collection, analysis, decision to publish, or preparation of the manuscript.

Data availability

The datasets generated and/or analyzed during the current study are not publicly available due to institutional policy on data confidentiality. They are available from the corresponding author upon reasonable request.

Declarations

Competing interests

All authors declare that they have no competing interests.

Ethical approval and consent to participate

This study received approval by the SingHealth Centralized Institutional Review Board (reference number CIRB 2018-2200) on 3/5/2018. It was registered on Clinicaltrials.gov (NCT03685422) with date of first registration on 24/09/2018. Written informed consent was obtained from every patient, and that this work was conducted in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.