從潛伏到重症：分枝桿菌感染對病人臨床結果的影響

Abstract

背景：分枝桿菌感染涵蓋由潛伏性結核感染（tuberculosis infection, TBI）至活動性結核（tuberculosis, TB）以及非結核分枝桿菌肺病（nontuberculous mycobacterial lung disease, NTM-LD）的一連串疾病譜。雖然全球結核防治已有進展，但潛伏感染治療不完全導致之再活化，以及重症病人中 NTM-LD 的日增負擔，仍對醫療體系構成挑戰。本論文整合臺大醫院整合醫療資料庫（National Taiwan University Hospital–Integrative Medical Database, NTUH-iMD）之兩項回溯性世代研究，探討預防與重症照護兩端之關聯，建立分枝桿菌疾病的連續性照護架構。 研究目的：第一部分分析 TBI 治療依從性與日後結核再活化之關聯；第二部分探討 TB 與 NTM-LD 在加護病房（intensive care unit, ICU）病人短期死亡率與脫離呼吸器預後的影響。兩者整合後，期能從預防與重症兩個角度提出整體分枝桿菌疾病管理的模式。 方法：第一研究納入 2016 至 2021 年於臺大醫院確診 TBI 或 TB 接觸者，依治療完成比例定義依從性，並以 Cox 模型分析結核再活化風險；對違反比例風險假設者，加入時間交互作用項以修正。第二研究納入 2006 至 2022 年間送檢呼吸道分枝桿菌培養之 ICU 病人，分為 TB 組、NTM-LD 組與陰性對照組。主要結果為 30 天死亡率，次要結果為 30 天呼吸器脫離存活率（ventilator-free survival, VFS）。多變項生存分析調整年齡、共病及 APACHE II 分數，並進行亞群及敏感性分析。 結果：TBI 族群共 1,432 人，治療完成者再活化風險降低 95%，依從性每增加 10%，再活化風險下降 23%，部分治療者仍具部分保護效果。ICU 族群共 5,996 人，TB 與 NTM-LD 皆與較差之 30 天預後相關。相較對照組，TB（aHR 2.33; 95% CI 1.92–2.83）與 NTM-LD（aHR 1.49; 95% CI 1.25–1.77）均增加死亡風險；NTM-LD 亦與較少呼吸器脫離天數及較低 30 天 VFS 有關（aHR 0.71; 95% CI 0.56–0.90）。結果顯示 NTM-LD 為影響重症病人預後的重要因子。 結論：本研究揭示從潛伏感染治療依從性到重症分枝桿菌感染預後之完整疾病連續體。提升 TBI 治療完成率可減少未來活動性結核病例，而及早辨識並個別化處理 ICU 之 TB 與 NTM-LD 患者，能改善短期生存。依從性、宿主脆弱性及診斷時效等因素跨越預防與治療兩端。整合數位依從性監測、快速分子診斷與醫院-公衛資料系統，可建立銜接性防治模式。本論文提出預防與重症照護整合的新典範，證明在分枝桿菌疾病控制中，橋接上游的公共衛生策略與下游的臨床照護，是邁向精準預防與結果導向醫療的重要方向。

Background: Mycobacterial diseases form a continuum from latent tuberculosis infection (TBI) to active tuberculosis (TB) and nontuberculous mycobacterial lung disease (NTM-LD). Despite progress in TB control, reactivation of incompletely treated latent cases and the rising burden of NTM-LD in critically ill patients remain major challenges. This dissertation integrates two retrospective cohort studies from the National Taiwan University Hospital Integrative Medical Database (NTUH-iMD) to examine upstream prevention and downstream critical-care outcomes. Objectives: To assess the relationship between adherence to TBI treatment and TB reactivation, and to evaluate the impact of TB and NTM-LD on short-term survival and ventilator outcomes in the intensive care unit (ICU). Together, these studies aim to bridge preventive and critical-care perspectives for comprehensive mycobacterial disease management. Methods: In the first study, individuals diagnosed with TBI or TB exposure (2016–2021) were identified from the NTUH-iMD and linked to national registries. Adherence was quantified as the proportion of doses completed and analyzed using Cox models, with time-by-covariate interactions applied when proportional hazards were violated. In the second study, ICU patients with respiratory cultures for mycobacteria (2006–2022) were classified as TB, NTM-LD, or no-growth controls. Primary and secondary outcomes were 30-day mortality and ventilator-free survival (VFS), respectively. Multivariable survival analyses adjusted for age, comorbidities, and APACHE II scores, with species-level subgroup and sensitivity analyses. Results: Among 1,432 TBI patients, complete treatment reduced TB reactivation by 95%, and each 10% increase in adherence lowered risk by 23%. Partial adherence provided moderate protection versus non-initiation. Among 5,996 ICU patients, both TB and NTM-LD predicted worse 30-day outcomes. Compared with controls, TB (adjusted hazard ratio [aHR] 2.33; 95% CI 1.92–2.83) and NTM-LD (aHR 1.49; 95% CI 1.25–1.77) were associated with higher mortality. NTM-LD also led to fewer ventilator-free days and lower 30-day VFS (aHR 0.71; 95% CI 0.56–0.90). These findings underscore NTM-LD as a clinically significant determinant of critical illness outcomes. Interpretation: The results delineate a continuum linking upstream adherence and downstream ICU survival. Strengthening TBI therapy adherence reduces future active TB, whereas early recognition and tailored management of NTM-LD and TB in the ICU improve short-term outcomes. Shared determinants—host vulnerability, diagnostic timeliness, and adherence—span prevention and treatment stages. Integration of digital adherence tools, rapid molecular diagnostics, and linked data systems can bridge public-health and hospital care. Conclusions: Bridging prevention and critical care is crucial for comprehensive mycobacterial disease control. Adherence to TBI therapy is a modifiable factor preventing reactivation, and NTM-LD represents an under-recognized cause of ICU mortality. This dissertation provides an evidence-based model connecting upstream prevention with downstream management, advancing precision prevention and outcome-driven care for mycobacterial infections.