Study Analyzes Large Language Model Dependence Factors Among Clinical Nurses

In a cross-sectional survey evaluating 459 clinical nurses across six hospitals in China, researchers found that more frequent use of large language models was significantly associated with higher scores in both functional and existential dependence, while receiving related training was associated with lower dependence scores. Functional items registered a higher overall mean score than existential items across the evaluated cohort. The investigation, conducted from August to September 2026 across three provincial-level administrative regions, assessed the current level of large language model dependence among clinical nurses and identified associated factors.

According to the authors, large language models are increasingly adopted across health-care and nursing environments to assist clinical nurses with retrieving information, improving administrative documentation, and supporting general clinical workflows. However, the researchers noted that inappropriate reliance on model outputs presents potential hazards during routine application. Unchecked dependence could diminish independent judgment and critical appraisal among nursing personnel, introducing potential risks to overall care quality and patient safety.

To measure dependence levels and evaluate contributing factors, the researchers deployed convenience sampling to recruit participants from six hospitals located across three provincial-level administrative regions in China. Data collection relied on a structured and validated survey instrument administered between August and September 2026. The analytical framework employed independent-samples t tests and one-way analysis of variance to evaluate variations in dependence scores across demographic characteristics, alongside multiple linear regression models designed to assess factors associated with functional and existential dependence.

Among the 459 valid responses gathered, the mean functional dependence score was 35.42 ± 5.81, and the mean existential dependence score was 27.18 ± 7.06. On a standardized 1 to 5 scale, these measurements corresponded to mean item scores of 3.94 ± 0.65 for functional dependence and 3.02 ± 0.78 for existential dependence. Univariate statistical evaluations demonstrated statistically significant variations in both functional and existential dependence scores based on participant age, educational attainment, frequency of model utilization, and receipt of related training, with all differences reaching significance at P < 0.05.

Multiple linear regression modeling confirmed that frequency of model usage and receipt of training remained significantly associated with both dependence categories after adjustment. More frequent utilization correlated with higher dependence scores, whereas prior training was linked to lower scores across both dimensions, while educational attainment displayed a more complex adjusted association. The authors emphasized that the observed scores do not establish validated severity cutoffs or define the clinical prevalence of problematic dependence, concluding that future nursing management should establish standardized training programs, institutional policies, and ethical oversight to guide cautious and responsible model use.

Subject of Research: Large language model dependence among clinical nurses

Article Title: Functional and existential dependence on large language models among clinical nurses and associated factors: a cross-sectional study

Article References: Ge, X., Li, X., Ni, Y., Wang, H., & Zhu, L. (2026). Functional and existential dependence on large language models among clinical nurses and associated factors: a cross-sectional study. BMC Nursing. https://doi.org/10.1186/s12912-026-05640-6

Image Credits: AI Generated

DOI: 10.1186/s12912-026-05640-6

Keywords: clinical nursing, large language models, artificial intelligence, nurse training, existential dependence, functional dependence, cross-sectional study, health care technology, Functional, existential, dependence, large