Establishment of HIV-negative neurosyphilis risk score model based on logistic regression

Abstract

Abstract Objective To establish the risk scoring model for HIV-negative neurosyphilis (NS) patients and to optimize the lumbar puncture strategy. Methods From 2016 to 2021, clinical information on 319 syphilis patients was gathered. Multivariate logistic regression was used to examine the independent risk factors in NS patients who tested negative for human immunodeficiency virus (HIV). Receiver operating characteristic curves (ROC) were used to assess the risk scoring model’s capacity for identification. According to scoring model, the timing of lumbar puncture was suggested. Results There were statistically significant differences between HIV-negative NS and non-neurosyphilis (NNS) patients in the following factors. These included age, gender, neuropsychiatric symptoms (including visual abnormalities, hearing abnormalities, memory abnormalities, mental abnormalities, paresthesia, seizures, headache, dizziness), serum toluidine red unheated serum test (TRUST), cerebrospinal fluid Treponema pallidum particle agglutination test (CSF-TPPA), cerebrospinal fluid white blood cell count (CSF-WBC), and cerebrospinal fluid protein quantification (CSF-Pro) (P < 0.05). Logistic regression analysis of HIV-negative NS patients risk factors showed that age, gender, and serum TRUST were independent risk factors for HIV-negative NS (P = 0.000). The total risk score (− 1 ~ 11 points) was obtained by adding the weight scores of each risk factor. And the predicted probability of NS in HIV-negative syphilis patients (1.6 ~ 86.6%) was calculated under the corresponding rating. ROC calculation results showed that the score had good discrimination value for HIV-negative NS and NNS: area under the curve (AUC) was 0.80, the standard error was 0.026 and 95% CI was 74.9–85.1% (P = 0.000). Conclusion The risk scoring model in this study can classify the risk of neurosyphilis in syphilis patients, optimize the lumbar puncture strategy to a certain extent, and provide ideas for the clinical diagnosis and treatment of HIV-negative neurosyphilis.

Keywords

• Treponema pallidum
• Neurosyphilis
• Logistic regression analysis
• Risk factors
• Risk score model