Relationship between nursing notes sentiment scores and length of hospital stay in elderly knee osteoarthritis patients undergoing total knee arthroplasty: a retrospective observational study.

Authors

Qingmei Wu, Qin Feng, Yiyang Liu, Wen Wang, Minyan Song, Fei Pan, Yufei Shao, Wei Zhang

Published in

BMC nursing. Jun 25, 2026. Epub Jun 25, 2026.

Abstract

With ongoing population aging, the prevalence of knee osteoarthritis (KOA) keeps increasing in the elderly population. For end-stage KOA patients, total knee arthroplasty (TKA) remains the definitive and effective treatment. Although numerous studies have examined factors associated with length of hospital stay (LOS) in TKA patients, these investigations predominantly focus on patient characteristics and surgical variables, often overlooking the "human factor" within the nursing process. This study aimed to examine the association between nursing note sentiment scores and prolonged length of hospital stay (PLOS) for KOA patients undergoing TKA.
In this retrospective observational study, sentiment analysis of nursing notes from the Medical Information Mart for Intensive Care-IV database was performed using natural language processing methods. Both random forest and AdaBoost algorithms were utilized to assess the significance of variables. Multivariable logistic regression was used to further clarify the relationship between PLOS and continuous/tertiles sentiment scores, and to construct a nomogram model. Quantile regression was utilized to determine dynamic associations between two continuous variables. The receiver operating characteristic curve was utilized to determine the nomogram model's performance.
648 KOA patients were included, with a median hospital LOS of 3.570 days. Multivariable logistic regression revealed that continuous scaled polarity was negatively associated with PLOS (OR = 0.209, 95%CI: 0.061-0.715). This association persisted when sentiment polarity was converted to a categorical variable (OR_Q2=0.589, 95%CI: 0.372-0.931; OR_Q3=0.616, 95%CI:0.386-0.982). Additionally, factors such as myocardial infarction, rheumatic disease, mild liver disease, smoking status, anticoagulant, and red blood cell distribution width were identified as influencing factors of PLOS (all OR > 1). The nomogram model constructed based on these factors demonstrated moderate discriminatory performance for PLOS (AUC: 0.702). Quantile regression indicated that when LOS < 70th percentile (3.750 days), LOS showed minimal correlation with scaled polarity. However, at LOS ≥ 70th percentile, LOS was significantly correlated with scaled polarity, with the negative correlation becoming stronger as LOS lengthened.
Sentiment polarity was negatively associated with hospital LOS, but this effect was primarily evident following PLOS. Monitoring nurses' sentiment states at critical junctures during inpatient treatment may enhance their motivation and facilitate patient recovery.
Not applicable.

PMID:
42351159
Bibliographic data and abstract were imported from PubMed on 26 Jun 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.