Implementation of Dragon Ambient Experience by Plastic Surgery Residents Improves Wellness and Decreases Burnout: A Pilot Study

Abstract

Background:

Burnout is a significant issue for plastic surgery residents, often worsened by administrative burdens such as clinical documentation. This study examined whether the Dragon Ambient Experience (DAX), a Microsoft artificial intelligence–powered tool integrated with Epic, can reduce documentation-related burdens and enhance resident wellness by increasing efficiency in completing clinic notes.

Methods:

In this pilot study, 6 plastic surgery residents received a 6-month DAX license. Participants completed presurveys assessing wellness, burnout, and job satisfaction using validated tools, including the National Aeronautics and Space Administration Task Load Index (NASA-TLX) and Mini-Z scales. They also recorded weekly hours spent on documentation outside clinic hours. Postsurveys were collected after 6 months to measure changes in these metrics.

Results:

At baseline, 2 out of 6 participants reported understanding care plans, and 3 out of 6 were able to give patients undivided attention. The mean Mini-Z score was 2.5 (moderate burnout), and the NASA-TLX workload score was 74.64 (high mental workload). Participants spent an average of 3.6 hours per week on documentation outside clinic hours. After DAX implementation, the NASA-TLX score decreased significantly to 51.94 (P = 0.04), and documentation time dropped to 0.75 hours per week (P = 0.02). The Mini-Z score slightly increased to 2.83; however, this was not statistically significant (P = 0.73).

Conclusions:

Our data highlighted documentation burdens on plastic surgery residents that significantly improved with the implementation of DAX, making it an important means of improving ambulatory workflow and resident wellness in plastic surgery.

Takeaways

Question: Can artificial intelligence (AI) software improve clinic flow and resident wellness?

Findings: The use of AI software significantly reduces the administrative burden, as evidenced by the time spent on clinic notes outside of clinic hours. Residents would recommend that others use this AI software. Residents reported significantly lower task load scores after using the AI software.

Meaning: Reduced documentation burden and taskload scores can improve resident wellness and decrease burnout.

INTRODUCTION

Burnout is a significant concern among healthcare providers. This condition is often accompanied by depression and substance abuse, contributing to a higher risk of errors in patient care.¹ As burnout can severely affect both physician wellness and patient outcomes, addressing its root causes is crucial. Resident physicians, who work long hours at base pay, are particularly vulnerable to burnout.^2,3 In a 2017 survey of Massachusetts physicians, those who spent more time on administrative tasks also reported higher burnout levels, with 67% stating that increased administrative burden reduced their ability to provide quality patient care. Among administrative duties, clinical documentation was identified as one of the most significant stressors.⁴

Burnout in plastic surgery residents is notably high, with rates reaching up to 57.5%.⁵ The excessive charting demands exacerbate feelings of emotional exhaustion and depersonalization, key components of burnout.⁶ A significant portion of this time is spent outside of regular working hours, as found in studies by Oxford et al⁷ and Maloney et al.⁸ As a result, reducing the administrative burden, especially related to clinical documentation, is a potential solution that may improve wellness and decrease burnout among residents.

Dragon Ambient Experience (DAX) is a Microsoft artificial intelligence (AI)–powered tool integrated into the Epic electronic health record (EHR) system. It automates the documentation of patient encounters, using natural language processing and machine learning to transcribe clinical notes in real time. This allows physicians to focus more on patient care rather than on administrative tasks. DAX is mobile-friendly, making it easily accessible in the clinical setting through Epic Haiku. Early studies have shown positive trends in reducing documentation time and enhancing provider engagement, with notable reductions in burnout among physicians using ambient AI technologies.^9–11

Despite the growing interest in AI tools such as DAX to alleviate the administrative load, there remains a lack of research focusing on resident outcomes. Most existing studies address attending physicians, highlighting a significant gap in the literature regarding the effects on residents.

Given the potential benefits of DAX in improving clinical efficiency and reducing administrative burden, our study aimed to evaluate its impact on plastic surgery residents. We hypothesize that the use of DAX will improve efficiency in note preparation, reduce burnout, and enhance overall wellness and job satisfaction.

We use 2 measures to assess the efficacy of DAX in reducing burnout in residents. First, the National Aeronautics and Space Administration Task Load Index (NASA-TLX), a validated tool used to assess perceived workload in clinical and cognitive settings.¹² In this study, we evaluated 3 domains: mental demand, effort required to reach personal satisfaction, and perceived time pressure. We also used the Mini-Z, a validated single-question survey that measures burnout, job satisfaction, and stress among healthcare providers.¹³ It offers a quick snapshot of provider wellness using a single-item burnout measure and related workplace factors. Ultimately, we anticipated that improving resident wellness would enhance long-term patient outcomes by fostering more engaged and fulfilled physicians.

METHODS

A pilot study was conducted by providing 6 plastic surgery residents with a 6-month license for DAX to be used in all clinic encounters. Before using the tool, residents completed a survey assessing their wellness, job satisfaction, and burnout, as shown in Supplemental Digital Content 1. (See survey, Supplemental Digital Content 1, which displays the pre-DAX survey. This survey was given to respondents before using DAX and included NASA-TLX and Mini-Z tools, https://links.lww.com/PRSGO/E474.) After 3 months of using DAX, participants completed a postsurvey. (See survey, Supplemental Digital Content 2, which displays the post-DAX survey. This survey was given to respondents after using DAX and included NASA-TLX and Mini-Z tools, https://links.lww.com/PRSGO/E475.). Changes in NASA-TLX, Mini-Z, and hours spent on documentation outside of clinic hours were the primary endpoints of interest. Secondary outcome measures included perceived quality of notes, satisfaction in ambulatory practice, the ability to give undivided attention to patients, whether providers felt patients could understand their care plans from the documentation, and qualitative reflections on DAX and overall wellness.

RESULTS

At baseline, 2 out of 6 participants reported that patients could understand their care plans from documentation, and 3 out of 6 stated they were able to give their patients undivided attention. (See figure, Supplemental Digital Content 3, which displays pre-DAX survey results. Two out of 6 respondents felt patients could understand care plans from their documentation [A], and 3 out of 6 felt they could give patients their undivided attention during the encounter [B], https://links.lww.com/PRSGO/E476.) The mean Mini-Z score was 2.83, indicating moderate burnout levels. The NASA-TLX workload score was 74.64, reflecting a high subjective mental workload. On average, participants spent 3.6 hours per week on documentation outside of clinic hours. Participants reported that if they had extra time in their day, they would use it for wellness activities, including exercise and time with friends or family. After wellness, residents most often cited the desire to use extra time for case preparation, studying for in-service examinations, and learning. A post hoc power analysis was conducted using the observed effect size (Cohen d = 1.64) and sample size (n = 6). With an α value of 0.05 and a 2-tailed paired t test, the estimated power was 87%, indicating adequate power to detect the observed difference in time spent on documentation despite the small sample size.

After implementing DAX for 6 months, the NASA-TLX workload score decreased significantly to 51.94 (P = 0.04), indicating a reduction in perceived mental workload. Participants also reported a significant decrease in time spent on documentation outside of clinic hours, with an average of 0.75 hours per week (P = 0.02). The mean Mini-Z score decreased to 2.5, but this change was not statistically significant (P = 0.73).

In terms of satisfaction, the median score for recommending DAX to a friend or colleague was 9. Additionally, 4 out of 6 participants felt that DAX improved the quality of their notes, and 4 out of 6 reported increased satisfaction in their ambulatory practice. On understanding care plans, 2 out of 6 participants reported improvement, and 5 out of 6 were able to give undivided attention to their patients. (See figure, Supplemental Digital Content 4, which displays the post-DAX survey results. Two out of 6 respondents felt patients could understand care plans from their documentation [A], and 5 out of 6 felt they could give patients their undivided attention during the encounter [B]. Four out of 6 felt that the use of DAX increased their satisfaction with their ambulatory practice, and 4 out of 6 felt that DAX improved the quality of their notes, https://links.lww.com/PRSGO/E477.)

DISCUSSION

The results of this pilot study suggest that the implementation of ambient AI technology such as DAX can meaningfully reduce the documentation burden and mental workload experienced by plastic surgery residents. Participants reported significantly decreased time spent on documentation outside of clinical hours and lower NASA-TLX scores, indicating improved cognitive efficiency. These improvements translated into a perceived increase in their ability to focus on patient care, give undivided attention during encounters, and feel more satisfied in the ambulatory setting. Although burnout scores showed only a modest and nonsignificant change, qualitative feedback emphasized that having more control over time creates the freedom to redirect it toward wellness and education. These findings underscore the potential for workflow-enhancing technologies to support wellness in meaningful, practical ways, even within a short time frame.

Our study expanded upon existing work in the field. Shah et al¹⁰ evaluated the impact of ambient AI scribe technology on physician burnout, task load, and usability in a pilot implementation at Stanford Health Care. The results showed significant reductions in task load and burnout, as well as improvements in usability and efficiency, suggesting that ambient AI scribes like DAX can enhance clinical workflows and reduce physician burden. The present study differs by focusing on the surgical field. This knowledge, until now, has been missing in the literature. By evaluating the impact of DAX on wellness and burnout in plastic surgery, the study expanded on previous research and demonstrated that ambient AI technology can be valuable not only in general medical practice but also in surgery.

Furthermore, there are no specific studies on how DAX influences burnout in resident physicians. Most of the existing data on DAX focus on attending physicians. For example, a cohort study by Haberle et al⁹ assessed the impact of DAX on provider engagement, time spent on electronic health record documentation, and other metrics among attending physicians, showing positive trends in provider engagement but no significant benefits to patient experience or productivity. As resident physicians bear the brunt of the administrative burden, we make a valuable contribution to the literature by examining how DAX can improve physician burnout at various stages of training.

Regarding cost, at our institution, physicians were quoted a group-discounted price of approximately $300 per user per month for DAX licenses; however, pricing can vary significantly depending on the size of the user base and institutional agreements. Although the upfront cost may be substantial, the potential for time savings in clinical documentation—as suggested by our findings—could translate into meaningful financial and workflow benefits. Future studies should rigorously evaluate the cost-effectiveness of ambient AI scribe technologies by quantifying documentation time saved, the impact on provider efficiency, and downstream effects on patient care and institutional resource use.

Although this study provided valuable insights into the impact of DAX on the wellness and burnout of plastic surgery residents, some limitations must be considered. The small sample size and single-institution design may limit the generalizability of the findings; however, this design was necessary to establish the validity of future investigations of how AI technology can be beneficial in plastic surgery and for resident physicians.

This study is also subject to potential bias from the Hawthorne effect, wherein residents may have altered their behavior simply due to awareness of being observed or participating in a study. Although participants were not explicitly informed of study hypotheses during implementation, the act of monitoring wellness and documentation habits may have influenced responses or behaviors independent of the intervention itself. Nonetheless, the magnitude of improvement observed supports the value of this pilot study in establishing a clear need for future research on DAX and its impact on resident wellness and efficiency.

Future studies with larger sample sizes and multi-institutional participation are recommended to enhance the generalizability of the findings. Longer follow-up periods and a control group that does not use DAX would also allow for a better understanding of the long-term impact of DAX on resident wellness, burnout, and job satisfaction. Furthermore, exploring the use of AI tools such as DAX across different medical specialties could provide insights into its broader applicability. Future research may also investigate the role of AI in medical education and clinical training environments, examining how it might support not only administrative tasks but also the development of clinical skills and decision-making in residents.

DISCLOSURE

The authors have no financial interest to declare in relation to the content of this article.