by A recent study published in JMIR AI examined the prevalence of anxiety and depression among healthcare workers (HCWs) in the United States during the COVID-19 pandemic. Leveraging machine learning methodologies, the study revealed the unique challenges faced by HCWs and provided insights into better supporting this crucial workforce.
Background:
Healthcare professionals often face higher rates of mental health issues, including depression, anxiety, and thoughts of suicide. The COVID-19 pandemic intensified the stress and workload of HCWs, as hospitals became overwhelmed with patients, operating beyond capacity. HCWs were forced to work longer hours under challenging conditions, such as shortages of equipment and resources, leading to difficult decisions in rationing care. Additionally, frontline workers were at higher risk of exposure to the virus, with limited access to essential protective gear. The strict quarantine guidelines also caused them to lose the support of social and familial networks.
Improved well-being and mental health support for HCWs are crucial for ensuring overall patient safety and a resilient healthcare system in the face of future disruptions.
About the Study:
The researchers analyzed the treatment transcripts of 820 HCWs who received digital psychotherapy from licensed providers between March and July 2020. To protect patient privacy, the transcripts were de-identified.
The HCWs included various healthcare providers like physicians, residents, nurses, social workers, and emergency medical service providers. All participants were self-referred and had active National Provider Identifiers (NPIs). The therapy was provided free of charge for one month through a telehealth platform that also treated non-HCWs.
To understand the unique challenges faced by HCWs, each provider was matched with a non-HCW based on symptoms, demographics, treatment start date, and state of residence. These non-HCWs were English-speaking US residents with internet access.
Prior to therapy, all patients underwent assessments for depression and anxiety by licensed providers using standardized measures.
The researchers used machine learning algorithms to analyze the treatment transcripts. Each HCW’s profession was identified from the transcript, and the text was processed to create a “vocabulary,” removing empty transcripts and infrequent words.
Results:
The majority of HCWs in the study were female (91%), with an average age of 31.3 years. New York State and California accounted for more than 25% of the sample. Nurses represented over half of the HCWs, while physicians accounted for less than 20%.
Approximately 35.2% of HCWs reported that this was their first experience with psychotherapy. Among the HCW patients, 56% were diagnosed with anxiety disorders, while 8.2% were diagnosed with depressive disorders. Prior to treatment, 73.3% of HCWs experienced either depression or anxiety.
Key topics frequently mentioned by HCWs included fears related to the coronavirus, work in intensive care units (ICUs) and hospital floors, masking and patients, and their specific roles. In contrast, non-HCWs mainly discussed pandemic anxiety and employer-related concerns.
Both HCWs and matched non-HCWs discussed topics related to mental health, such as panic attacks, mood disturbances, and grief experiences. HCWs also frequently mentioned disruptions to their sleep.
Providers with moderate to severe depression or anxiety were more likely to discuss hospitals or specific areas like ICUs. They were also more likely to mention mood alterations or disruptions in sleep compared to matched controls.
Conclusion:
Using machine learning techniques, this study compared the experiences of 820 HCWs with 820 matched non-HCW patients undergoing therapy on the same platform. The findings indicate that HCWs faced unique challenges associated with their work during the COVID-19 pandemic, exacerbating their existing stress levels. This emphasizes the need for prioritizing the mental health of healthcare workers.
The study has certain limitations, including the lack of inclusion of individuals with limited access to virtual therapy and the skewed representation of female nurses in the sample. Future research could explore more complex linguistic models and include non-English transcripts.
Nevertheless, this study provides actionable insights into the challenges faced by HCWs during the pandemic. It also demonstrates the utility of machine learning algorithms in processing and analyzing large datasets while maintaining participant privacy.
*Note:
1. Source: Coherent Market Insights, Public sources, Desk research
2. We have leveraged AI tools to mine information and compile it
