Department of Statistics, College of Natural and Computational Sciences, Dilla University, Dilla, Ethiopia.
Department of Statistics, College of Natural and Computational Sciences, Mattu University, Mattu, Ethiopia.
Sci Rep. 2024 Oct 13;14(1):23912. doi: 10.1038/s41598-024-73209-x.
Globally, the prevalence of multidrug-resistant tuberculosis (MDR-TB) has been increasing recently. This is a major public health concern, as MDR-TB is more difficult to treat and has poorer outcomes compared to drug-sensitive tuberculosis. The main objective of the study was to identify risk factors for recurrent multidrug-resistant tuberculosis, at Alert Specialized Hospital, Addis Ababa, by using different parametric shared frailty models. From January 2016 to December 2021, a retrospective study was conducted on MDR-TB patients at Alert Specialized Hospital in Addis Ababa. The data for the study were collected from the medical records of MDR-TB patients at the hospital during this time period. Gamma and inverse-Gaussian shared frailty models were used to analyze the dataset, with the exponential, Weibull, and lognormal distributions included as baseline hazard functions. The data were analyzed using R statistical software. The median recurrence time of the patients was 12 months, and 149 (34.3%) had recurrences. The clustering effect was statistically significant for multiple drug-resistant tuberculosis patients' recurrence. According to the Weibull-Inverse-Gaussian model, factors that reduced time to MDR-TB recurrence included lower weight (ɸ = 0.944), smoking (ɸ = 0.045), alcohol use (ɸ = 0.631), hemoptysis (ɸ = 0.041), pneumonia (ɸ = 0.564), previous anti-TB treatment (ɸ = 0.106), rural residence (ɸ = 0.163), and chronic diseases like diabetes (ɸ = 0.442) were associated with faster recurrence. While, higher education (ɸ = 3.525) and age (ɸ = 1.021) extended time to recurrence. For weight increment, smokers and alcohol users, clinical complications of hemoptysis and pneumonia, patients with pulmonary disease who had a history of previous anti-TB treatment, and being rural residents are prognostic factors. There was a significant clustering effect at the Alert Specialized Hospital in Addis Ababa, Ethiopia. The Weibull-Inverse Gaussian Shared Frailty Model was chosen as the best model for predicting the time to recurrence of MDR-TB.
全球范围内，耐多药结核病（MDR-TB）的患病率最近一直在上升。这是一个主要的公共卫生关注点，因为与敏感结核病相比，MDR-TB 更难治疗，且预后更差。本研究的主要目的是使用不同的参数共享脆弱性模型，确定在亚的斯亚贝巴 Alert 专科医院复发性耐多药结核病的危险因素。2016 年 1 月至 2021 年 12 月，对亚的斯亚贝巴 Alert 专科医院的耐多药结核病患者进行了回顾性研究。本研究的数据来自该医院在此期间的耐多药结核病患者的病历。使用伽马和逆高斯共享脆弱性模型来分析数据集，其中包括指数、威布尔和对数正态分布作为基线危险函数。数据分析使用 R 统计软件进行。患者的中位复发时间为 12 个月，有 149 人（34.3%）复发。多重耐药结核病患者的复发存在统计学显著的聚类效应。根据威布尔-逆高斯模型，降低耐多药结核病复发时间的因素包括较低的体重（ɸ = 0.944）、吸烟（ɸ = 0.045）、饮酒（ɸ = 0.631）、咯血（ɸ = 0.041）、肺炎（ɸ = 0.564）、以前的抗结核治疗（ɸ = 0.106）、农村居住（ɸ = 0.163）和糖尿病等慢性病（ɸ = 0.442）与更快的复发有关。而较高的教育程度（ɸ = 3.525）和年龄（ɸ = 1.021）则延长了复发时间。对于体重增加、吸烟者和饮酒者、咯血和肺炎的临床并发症、以前有抗结核治疗史的肺部疾病患者以及农村居民来说，这些都是预后因素。在埃塞俄比亚亚的斯亚贝巴的 Alert 专科医院存在显著的聚类效应。威布尔-逆高斯共享脆弱性模型被选为预测耐多药结核病复发时间的最佳模型。
