Application value of serum protein indicators in constructing the early prediction model for the prognosis of patients with pulmonary tuberculosis

Abstract

Objective: To analyze the clinical significance of laboratory examination indicators as the key prognostic factors and to construct an early prediction model for prognosis assessment of pulmonary tuberculosis patients. Methods: The basic information, biochemical indexes and blood routine items of 163 tuberculosis patients (144 males and 19 females, aged 41-70 years, with an average age of 56 years) and 118 healthy persons who underwent physical examination (101 males and 17 females, aged 46-64 years, with an average age of 54 years) in Suzhou Fifth People's Hospital from January 2012 to December 2020 were retrospectively collected. According to the presence of Mycobacterium tuberculosis after six months of treatment, the enrolled patients were divided into a cured group (96 cases) and a treatment failure group (67 cases). To analyze the baseline levels of laboratory examination indicators between these two groups, we screened the key predictors and the binary logistic regression method in SPSS statistics software was used to construct the prediction model. Results: The baseline levels of total protein, albumin, prealbumin, glutamic-pyruvic transaminase, erythrocyte, hemoglobin and lymphocyte were significantly higher in the cured group than in the treatment failure group. After 6 months of treatment, the indexes of total protein, albumin and prealbumin increased significantly in the cured group, but remained at the low levels in the treatment failure group. Receiver operating characteristic (ROC) curve analysis showed that total protein, albumin and prealbumin as independent predictors for forecasting the prognosis of pulmonary tuberculosis patients had the highest prediction accuracy. Logistic regression analysis showed that the combination of these three key predictors could construct the best early prediction model for assessing the prognosis of pulmonary tuberculosis patients, with a prediction accuracy of 0.924 (0.886-0.961), sensitivity of 75.0%, specificity of 94%, showing an ideal prediction accuracy. Conclusions: The routine test indexes of total protein, albumin and prealbumin show good application value in the construction of early prediction model for prognosis evaluation of pulmonary tuberculosis treatment. The combined prediction model consisting of total protein, albumin and prealbumin is expected to provide a theoretical basis and reference model for precision treatment and prognosis assessment of tuberculosis patients.