Multivariate Analysis Of Factors Influencing Length Of Hospital Stay In Patients With Rheumatoid Arthritis

Objective:This study retrospectively investigated the influencing factors of Length of stay(LOS)in patients with rheumatoid arthritis(RA),and the predictive value of hospitalization information for LOS.To provide a scientific basis for reasonable control of LOS in RA patients,optimize hospital efficiency,and improve treatment effect.Method:This study was a retrospective study.From the medical data intelligent management platform,patients with a clear main diagnosis of rheumatoid arthritis in the First Hospital of China Medical University from January 1,2019 to December 31,2021 were collected.After consulting the electronic medical records,the demographic characteristics,admission information,clinical laboratory serological indicators,comorbidities and drug use during hospitalization of the patients were extracted and the database was established.Generalized linear model was used to explore the independent influencing factors of LOS in RA patients.The data set was randomly split into a training set and a validation set at a ratio of 7:3,and a binary Logistic regression model was constructed in the training set according to whether the hospitalization days of RA patients were more than 7 days as the dependent variable.Receiver operating characteristic(ROC)curve,area under the curve(AUC),calibration curve,C curve)and Hosmer-Lemeshow(H-L test)were used to evaluate the performance of the model,and the validation set was used to test the performance of the model.Results:A total of 1150 subjects were included in this study.(1)the analysis on the factors affecting the hospitalization time,single factor analysis results showed that age,occupation,payment,admission time,joint swelling,joint pain,28 joint Disease Activity Score(diseases Activity Score 28,DAS28),Erythrocyte sedimentation rate(ESR),Red cell distribution width(RDW),Immunoglobulin M,Ig M),carbohydrate Antigen 125(Carbohydrateantigen 125,CA125),carcinoembryonic Antigen(Carcinoma Embryonic Antigen,CEA),rheumatoid vasculitis,hypertension,autoimmune disease,non-steroidal anti-inflammatory drugs,and glucocorticoids were significantly different from LOS in RA patients.(2)In the analysis of influencing factors of length of stay,multivariate analysis showed that the independent risk factors for LOS in RA patients were:abnormal RDW level(β=0.340,P=0.010);Abnormal CEA level(β=0.555,P=0.014);DAS28 was highly active(β=2.676,P<0.001).DAS28was moderately active(β=1.437,P<0.001).DAS28 was in the low activity stage(β=1.172,P=0.003).Rheumatoid vasculitis(β=1.273,P=0.006);The use of non-steroidal anti-inflammatory drugs during hospitalization(β=0.407,P=0.001);The use of glucocorticoids during hospitalization(β=0.259,P=0.043);The independent protective factor for LOS in RA patients was self-payment(β=-0.995,P<0.001).(3)In the prediction model(LOS>7 days),based on binary logistic regression analysis,it was found that self-paid(OR:0.414,95%CI:0.189-0.904,P=0.027);DAS28 high activity OR:2.785,95%CI:2.035-3.089,P<0.001).RDW(OR:2.605,95%CI:1.297-5.233,P<0.001);Ig M(OR:2.605,95%CI:1.297-5.233,P=0.007)constituted the prediction model.(4)The AUC value of the training set was 0.664,the cut-off value was 0.380,the sensitivity was 73.9%,and the specificity was 52.4%.The AUC value of the validation set was 0.648,the cut-off value was 0.862,the sensitivity was 81.5%,and the specificity was 44.4%.In the H-L test,χ~2=3.447,P=0.903in the validation set.A total of 344 patients were included in the validation set,and the median length of hospital stay was 9(8-10)days.A total of 225 people had consistent prediction results,with an accuracy rate of 65.4%,including 36 true negatives,189 true positives,30 false positives and 83 false negatives.Conclusion:(1)The influencing factors of LOS in RA patients were as follows:payment method,combined rheumatoid vasculitis,RDW,CEA,DAS28,taking non-steroidal anti-inflammatory drugs and glucocorticoids during hospitalization.(2)RA patients with medical insurance payment,high activity of DAS28,higher RDW and higher Ig M have a higher risk of LOS>7 days.The above factors constitute a predictive model for LOS in RA patients.The model has certain discrimination ability and high calibration ability,and has a moderate accuracy in the validation set.It has certain clinical application value.