Evaluation Of New Prediction Model On Prognosis Of Community-acquired Pneumonia

Objective: Community-acquired pneumonia(CAP)is the most common respiratory parenchymal pulmonary inflammation in the world.Evaluation of the severity of pneumonia is an important part of the diagnosis and treatment of pneumonia.In 2016,the guidelines for community-acquired pneumonia in China clearly proposed the evaluation standard of CURB-65 and is widely used in clinical practice.But for patients with immunodeficiency,the prediction model cannot be used.By the increasing number of CAP patients with low immune function and underlying diseases in the clinic,how to better determine their prognosis has been a question for clinicians.Some scholars have proposed the model of CURB-65 L.The factor of decreased lymphocyte count has certain reference value.This study used retrospective analysis to explore the effects of commonly used scoring systems combined with serum biomarkers on the prognosis of community-acquired pneumonia,forming a new prediction model to evaluate the effect of the new prediction model on the prognosis of CAP,aiming to bring some guiding significance to clinical diagnosis and treatment.Methods: A retrospective analysis was performed to collect 201 cases of CAP patients who attended the Department of Respiratory and Critical Care Medicine,the second affiliated hospital of Anhui Medical University from January 2015 to December 2018.The general clinical characteristics of the patients,such as gender and age,body temperature,blood pressure,respiratory rate,mental state and comorbidities,etc were analyzed.The patient's venous blood was collected before treatment with antibiotics,and serum biochemical indicators(white blood cell count,neutrophil count,lymphocyte count,procalcitonin,C-reactive protein,blood urea nitrogen,etc.).The CURB-65,A-DROP,q-SOFA scores and related biological indicators of the patients were calculated,and perform a descriptive analysis,taking the 30-day prognosis(cured or died)of the patient as the endpoint of clinical observation,draw the ROC curve of four scoring systems and each biomarker,calculate the AUC value,and compare its effectiveness in predicting the 30-day mortality of CAP patients.The scoring system with better predictive power is combined with biological indicators to form a new prediction model and compare the predictive value of the new prediction model.Results: During the study period,a total of 201 adult CAP patients were included.The median age of patients included was 66(51-77)years,and 64.7% were male.The top three major comorbidities are chronic heart failure(31.0%),chronic obstructive pulmonary disease(COPD)(25.4%),and diabetes(19.4%).The 30-day total mortality rate is 12.9%,and the rate of admission to the ICU is 27.9%.The average hospital stay for all patients was 12(7-18)days.The CURB-65,q-SOFA,A-DROP and NLR all have predictive value for 30-day mortality in patients with CAP.The area under the curve(AUC)is 0.73(0.63-0.83),0.63(0.60-0.74),0.69(0.61-0.81)and 0.69(0.60-0.80)respectively.Among the biological indicators,lactate dehydrogenase(LDH),procalcitonin(PCT),albumin(Alb)and lymphocyte counts have predictive value for 30-day mortality in patients with CAP.The areas under the curve for mortality were 0.61(0.50.0.73),0.86(0.77,0.94),0.71(0.63,0.81)and 0.76(0.66,0.85).The PCT and lymphocyte counts were combined with the CURB-65 scoring system to form CURB-65 P,CURB-65 L,and CURB-65 LP.The areas under the curve for mortality were 0.80(0.73-0.89),0.79(0.70-0.87)and0.84(0.77-0.91).Conclusion: Among the new prediction models,CURB-65 LP has the best predictive performance,with sensitivity and specificity of 81% and 68%,respectively.Among the common scoring systems,CURB-65 is slightly more effective in predicting 30-day mortality than the other three systems.Its sensitivity and specificity are 74% and 65%.The predictive power of PCT and lymphocyte count are slightly higher,with sensitivity and specificity of 81% and 90%,71% and 77%,respectively.