Virtual Pet–Assisted Therapy to Alleviate Symptoms of Long COVID: A Prospective Pilot Interventional Study

Introduction

COVID-19 rapidly developed into a global pandemic within just a few months after the first case was reported in Wuhan, China, in early December 2019 [1]. In Japan, the first case was confirmed on January 15, 2020, and infections have continued to spread since then. Among these challenges are the post-acute sequelae of COVID-19, known as post–COVID-19 condition and commonly referred to as “long COVID,” which are characterized by persistent symptoms that extend beyond the infectious phase [2]. The definition of long COVID has remained provisional, with no established standard. In June 2024, the National Academy of Sciences, Engineering, and Medicine released a new definition to support consistent diagnosis, documentation, and treatment of long COVID, stating that “Long COVID is a chronic disease associated with SARS-CoV-2 infection characterized by persistent, recurrent, or progressive symptoms affecting one or more organ systems that persist for at least three months” [3]. Symptoms include shortness of breath, cough, persistent fatigue, fatigue after exertion, difficulty concentrating, memory loss, recurring headaches, dizziness, increased heart rate, sleep disorders, taste or smell problems, bloating, constipation, and diarrhea, among others. Several studies on pharmacological treatments for long COVID are currently underway. Corticosteroids, colchicine, and pirfenidone are expected to be effective for respiratory symptoms [4], while probiotics are thought to alleviate fatigue by influencing short-chain fatty acids involved in regulating inflammatory responses [5]. These findings suggest several potential treatment options. However, each trial is ongoing or has limitations, such as small sample sizes; therefore, the widespread use of these medications in clinical practice has not yet been achieved. In Japan, symptomatic therapies or traditional Japanese (Kampo) medicine are currently the primary approaches used to treat symptoms. Currently, there is no established standard treatment for long COVID [6]. However, due to its diverse symptoms, long COVID is recognized as a multisystem disease characterized by symptoms affecting the respiratory, hematological, immune, cardiovascular, and neuropsychiatric systems, either individually or in combination [7]. Pharmacological therapy is just one aspect of treatment for long COVID; a multidisciplinary, holistic approach that incorporates non-pharmacological interventions, such as cognitive behavioral therapy, counseling, and rehabilitation [8], is recommended [4,7]. Long COVID significantly reduces patients’ quality of life and often limits their ability to work [9,10], making it a serious social issue. Manuel et al reported that many of the 416 long COVID patients in their study experienced adverse occupational effects (41.0%), economic losses (40.5%), and changes in their stress coping abilities (87.5%). They also encountered changes in their work life, such as taking sick leave, reduced income, shorter working hours, or an inability to work [11]. Additionally, Ziauddeen et al reported that 37.6% of all study participants experienced a loss of income [12]. Under these circumstances, establishing effective treatments for long COVID is imperative. Exploring effective non-pharmacological therapies alongside pharmacological treatments is essential for developing a robust treatment strategy.

As a non-pharmacological therapy, we propose virtual pet–assisted therapy (VAT). VAT is an innovative treatment that substitutes the non-pharmacological approach of animal-assisted therapy (AAT) with virtual reality (VR) technology. AAT is aimed at improving mental and physical health through interaction with animals. It was first introduced as part of clinical care in the United States in the 1970s [13], subsequently gaining recognition in Japan. Several studies have been conducted on the efficacy of AAT. One study showed that, in elderly residents of long-term care facilities with mental illnesses such as depression and dementia, AAT led to a 50% improvement in the Geriatric Depression Scale, and the average score on the Mini Mental State Examination more than doubled in the AAT group, compared with the control group [14]. In Japan, AAT has been reported to provide physiological, psychological, and social benefits [15] and has been implemented in clinical practice in some medical institutions and eldercare facilities. Given these findings, AAT is expected to be effective in treating the psychological symptoms of long COVID, such as sleep disturbances and memory impairments.

However, infection control measures during the COVID-19 pandemic minimized interspecies contact, making it difficult to implement AAT using live animals within hospitals. Therefore, we explored VR technology as an alternative. VR technology creates simulated, interactive 3-dimensional environments that users can engage with through head-mounted displays. It has made remarkable progress in recent years and has garnered attention for its potential contributions to healthcare. VR therapy may be useful in delivering cognitive behavioral therapy, and its immersive nature may contribute to symptom relief. In the field of psychiatry, conventional treatments have mainly been limited to interpersonal psychotherapy and pharmacotherapy. However, VR, by providing diverse stimuli [16], is drawing attention as a new therapeutic option. For example, VR therapy has been shown to reduce depression severity and self-criticism in patients with depression [17]. Clinical studies have also shown the use of VR in psychiatric conditions such as schizophrenia, autism spectrum disorder, and dementia [18]. On the other hand, in patients with chronic obstructive pulmonary disease, VR has been utilized in pulmonary rehabilitation, alleviating psychological symptoms such as depression and anxiety while also improving exercise tolerance related to physical symptoms such as dyspnea [19,20]. Additionally, vestibular rehabilitation using VR for patients with dizziness caused by peripheral vestibular dysfunction showed the same efficacy as conventional therapies, along with a significant improvement in patient satisfaction [21]. Similar to AAT, VR therapy is effective for psychological symptoms and may also address physical symptoms such as dyspnea and dizziness.

Many patients with long COVID experience physical and psychological distress. Despite the growing body of research supporting the potential benefits of AAT and VR therapy in healthcare, reports on its effectiveness and feasibility for patients with long COVID remain extremely limited [22], and studies involving the use of VAT in long COVID are even more lacking. Therefore, we hypothesized that VAT could improve long COVID symptoms. The aim of this clinical study was to test this hypothesis and evaluate the effect of VAT on specific long COVID symptoms.

Material and Methods

DESIGN AND PARTICIPANTS:

Following the COVID-19 pandemic, the Department of General Medicine at Juntendo University Hospital established a long COVID outpatient clinic on October 1, 2021, targeting patients whose symptoms persisted for at least 2 months following COVID-19 onset, specifically patients whose symptoms were present 3 months after infection. Patients visit the clinic once every 1 to 2 months and primarily receive pharmacological treatment centered around traditional Japanese (Kampo) medicine.

This prospective interventional study was open-label and involved patients aged 18 years and older attending the clinic, regardless of symptom severity, who provided written informed consent after receiving a full explanation and demonstrating adequate understanding. Patients were excluded if they had vertigo severe enough to impair their standing ability, claustrophobia, or communication difficulties due to cognitive decline. Furthermore, this study was limited to patients who meet the previously mentioned criteria for long COVID, and patients with sequelae from COVID-19 vaccination were excluded. The study spanned 5 months (July 31, 2023, to December 12, 2023). VAT was administered to all consenting patients prior to their clinic consultation. To minimize the risk of VR-induced motion sickness and avoid disrupting clinic workflow, the VAT session was limited to 10 minutes. Participants completed a brief 5-minute questionnaire before and after the session.

VARIABLES:

Basic background data collected included age, sex, COVID-19 vaccination history and number of doses, number of symptoms [23,24], and number of visits to the long COVID outpatient clinic. Additionally, participants were asked about past and current dog ownership.

As outcome variables, symptoms associated with SARS-CoV-2 infection were referenced from 2 previous reports [23,24]. Of the 42 total symptoms listed, those that could potentially improve in a short period were independently assessed by R.K. and T.M., achieving a concordance rate of 92%. Discrepancies were resolved by M.S., who decided which symptom would be included. Ultimately, the analysis was limited to 12 symptoms considered potentially improvable through VAT: fatigue, generalized pain, muscle pain, limited movement, abdominal symptoms, dyspnea, olfactory dysfunction, gustatory dysfunction, sleep disturbance, memory impairment, tinnitus, and dizziness. Participants rated the severity of each symptom on a 10-point scale before and after VAT. As this is a pilot study, this rating method was used instead of validated instruments due to its simplicity, patient-friendliness, and suitability for a brief clinical assessment.

ETHICS:

This study was approved by the Ethics Committee of Juntendo University (approval No. E23-0078, UMIN ID: UMIN000051253). Participants received written information outlining the study purpose, expected benefits, and potential risks, after which they provided written informed consent. If participants chose to withdraw from the study or revoke their consent, they could leave at any time. When handling samples associated with the study, we assigned research subject codes that were unrelated to participants’ personal information and took necessary precautions to protect the confidentiality of research participants.

INTERVENTION:

VAT was conducted using a technology that allowed participants to interact within a VR environment with a virtual dog, referred to as a “virtual pet” (Figure 1), which was developed by Kenta and Mihoko [25]. Participants wore a head-mounted display device – specifically, the HTC Vive Pro headset (HTC Corporation, New Taipei City, Taiwan) (Figure 2) – and engaged in a 10-minute session prior to their clinical consultation, during which they were free to interact with the virtual pet in a VR environment designed to resemble a grassy field (Figure 3). The VR environment was created using Unity, a VR game development software engine. Using the device’s accompanying controllers, participants could perform various interactive actions (Figure 4). Participants extended their hands toward the approaching dog and stroked it, while the controller transmitted vibrations to their hands, simulating the sensation of actually petting the dog. By pressing a button on the controller, participants could grasp a ball regardless of distance and throw it far away. The virtual pet would chase the ball and return it to the participant. Based on findings from animal behavior science [26], these interactions were designed to elicit attachment behaviors from the virtual pet toward the participant. The more time participants spent with the virtual pet and the more they interacted, the more affectionate behaviors the virtual pet would display toward them. This dynamic allowed participants to experience an interaction that closely resembled real-life engagement with an actual dog. During the VAT implementation, staff were always available to assist with operating the equipment and ensuring a safe environment.

STATISTICS:

All statistical analyses were performed using JMP version 18.0 (SAS Institute, Cary, NC, USA). All reported P values were 2-tailed, and a P value of less than 0.05 was considered statistically significant. For continuous variables, results are presented as median and interquartile range (IQR) for non-normally distributed data, and as mean±standard deviation (SD) for normally distributed data. Categorical variables are presented as numbers and percentages. Since the symptom severity scores before and after the intervention were not normally distributed, the Wilcoxon signed-rank test was used for comparison. For symptoms with statistically significant differences, effect sizes were calculated. Multiple regression analysis was performed with the dependent variable being the difference in total severity scores before and after treatment, while age, sex, and number of vaccine doses served as covariates.

Results

A total of 67 patients participated in the study, of whom 25 were excluded owing to missing data (n=5), being outside the age criteria (n=2), and sequelae from COVID-19 vaccination (n=18), resulting in a final analysis cohort of 42 participants (Figure 5). Most of the participants were women (71.4%, n=30) and had received COVID-19 vaccination (85.7%, n=36). The majority previously owned a dog (64.3%, n=27), while few were current dog owners (19%, n=8). The median age of the participants was 46 (IQR: 34–50) years. The median number of vaccine doses received were 2 (IQR: 2–3), symptoms experienced, 5 (IQR: 3–7), and outpatient visits at the time of the first VAT session, 7 (IQR: 4–10) (Table 1).

The mean total symptom severity score decreased from 34.6±18.3 before VAT to 32.1±18.9 after VAT (P=0.004) (Figure 6). For individual symptoms, fatigue score decreased from 6.3±2.6 to 5.7±2.9 (P=0.004); dyspnea from 2.3±3.0 to 1.9±2.8 (P=0.038), memory impairment from 4.8±3.0 to 4.2±3.2 (P=0.015), and tinnitus from 2.3±2.9 to 1.8±2.5 (P=0.012). No significant improvement was observed in the other 8 symptoms. The effect sizes were as follows: the total symptom severity score was 0.13, fatigue was 0.22, dyspnea was 0.14, memory impairment was 0.19, and tinnitus was 0.19 (Table 2).

A multiple regression analysis was conducted using the difference in the total severity score of all symptoms as the dependent variable; however, patient factors such as age, sex, and the number of vaccine doses did not affect the total symptom severity score (Table 3).

Discussion

LIMITATION:

This study has several limitations. First, several patient background factors, such as disease duration, comorbidities, and the severity of long COVID, were lacking, which may have introduced bias in interpreting the results for specific groups. Additionally, since this is a pilot study, no similar studies are available, and sample size and power analysis were not calculated. Regarding symptoms that improved significantly, the effect sizes were all low to moderate, so it cannot be said that the intervention was very effective.

Second, the duration of the VAT intervention was short and probably insufficient to adequately assess its effects. To avoid VR-related motion sickness and ensure smooth outpatient examinations, the intervention duration was limited to 10 minutes. In contrast, some studies have implemented 30-minute VR sessions for patients with long COVID [22], which resulted in a passive engagement. Moving forward, it will be essential to explore effective and safe intervention durations by evaluating and comparing various lengths while being mindful of potential adverse effects.

Third, since the intervention was administered only once, the sustainability of the VAT effect remains unknown. Long COVID symptoms are characterized by their long-term persistence, and treatment effects must also be sustained. Future studies should involve follow-up symptom evaluations and interventions after a certain period from the first intervention to assess the long-term effects of VAT.

Fourth, symptom severity assessment before and after VAT was based on subjective self-reports only. For instance, fatigue could be quantitatively assessed using the Fatigue Assessment Scale or other relevant questionnaires tailored to specific symptoms. The 10-point scale used in this study is familiar to Japanese individuals from their school days and was chosen for its clarity. However, its reliability has not been verified.

Fifth, this study was conducted at a single facility in Japan, leading to selection bias and limiting the generalizability of the results. Cultural attitudes toward VR, AAT, and technology-based interventions vary by country and region. In particular, the emotional significance and symbolic value of dogs differ greatly based on cultural norms, religious beliefs, and personal experiences. For example, participants from cultures that traditionally view dogs as unclean or do not emphasize the bond between humans and animals may not experience the same emotional benefits from interacting with virtual dogs. Similarly, individuals without prior pet ownership experience or affinity may be less responsive to VAT or struggle to engage meaningfully within the virtual environment.

Sixth, individual differences such as past pet ownership experience, technical literacy, age, personality traits, and psychological readiness may also influence the effectiveness of VAT. In this study, we collected some information on past or current dog ownership, but we did not systematically evaluate factors such as affinity for animals, cultural adaptability to virtual environments, and digital accessibility. These variables may affect the results and should be examined in future studies using stratified analysis or mixed methods.

Seventh, the absence of a control group limited our ability to accurately evaluate the impact of the intervention. The placebo effect, participants’ expectations, or the mere act of engaging in a new activity may have contributed to the observed changes.

Finally, the VAT implemented in this study required not only VR equipment but also gaming computers and multiple large machines that necessitated specialized knowledge to operate. For future clinical use in medical institutions or homes, it will be essential to miniaturize the devices and simplify their operation.

Despite these limitations, this study revealed significant effects on the particularly challenging symptoms of long COVID, such as fatigue and brain fog, for which treatment options have not been established. These results provide initial evidence suggesting that VAT may be an effective adjunctive therapy for long COVID. To ensure fair and meaningful outcomes across diverse patient populations, future studies must adopt culturally sensitive and individualized approaches.

Conclusions

VAT has demonstrated potential efficacy in alleviating multiple symptoms, including fatigue, which is one of the most representative symptoms of long COVID attributed to psychological factors. This result is consistent with our hypothesis that VAT could provide relief. However, the pilot study with an uncontrolled design prevents us from concluding that VAT was the cause of this change. To assess its efficacy, randomized controlled trials with an active control group (such as a non-interactive VR environment) are necessary across multiple session protocols and must be evaluated using validated outcome measures. Combining VAT with traditional treatments may offer a comprehensive approach to addressing both the physical and psychological aspects of long COVID. Additionally, if implemented in clinical practice, VAT will enable consistent and effective delivery of treatment for patients with long COVID without depending on medical staff.