老年族群感染症：流行病學、長期感染軌跡模型與相關之不良結果

Abstract

[研究背景] 感染症是老年人最重要的死亡原因之一，除了造成老年人短期內的疾病負擔外，亦可能與長期的不良結果相關。然而，過往文獻中針對老年人感染症的流行病學資料及疾病負擔尚不全面，對於感染症與臨床不良結果的相關性亦是著眼於單一感染事件的長期影響，而未能縱貫性探討感染事件重複暴露的累積效應，可能影響對於老年人感染事件與不良結果相關性的探討。 [研究目的] 估算臺灣老年人發生感染事件之全國發生率，以及重複感染事件對於疾病負擔的影響。追蹤老年人感染症發生情形，並以群組化軌跡模型分析老年人感染事件發生情形存在哪些不同的長期軌跡模式，探討不同感染事件軌跡老年人之特性差異，及不同感染事件軌跡模式與後續臨床不良結果之相關性。 [研究方法] 本研究第一部分屬於回溯性橫向研究設計，利用臺灣全民健保資料庫蒐集 2000 年至 2019 年 65 歲以上老年人發生住院感染事件之紀錄，計算年度發生率，並以性別、年齡組進行分層比較，發生率之計算僅列入每年每人第一筆相關住院感染事件，並以 Joint regression model 分析發生率變化趨勢，以 Poisson regression 比較各分層的發生率差異。本部分研究亦利用住院期間之醫療紀錄比較重複發生感染事件之老年人，其初次住院感染事件與後續重複感染事件的間隔天數、臨床不良結果與醫療利用差異。 本研究第二部分屬於回溯性世代研究設計，蒐集臺灣全民健保資料庫中於 2011 年曾有住院紀錄之感染事件的 65 歲以上老年人，並追蹤其 2011 年第一次住院感染事件之入院日起五年內的感染事件發生情形。記錄完整存活於五年軌跡觀察期的研究對象是否曾於軌跡觀察期內發生住院感染事件，並以每三個月為時間單位，以群組化軌跡模型將軌跡進行分組。分組完成後，比較各軌跡組別之基本特性與共病差異，並定義不良結果追蹤期為軌跡觀察期結束後五年，納入研究對象進入不良結果追蹤期前一年之共病情形與用藥情形做為校正變項，並使用Kaplan-Meier curve 及 Cox proportional hazard model 探討不同軌跡組別與發生全死因死亡的相關性。 [研究結果] 本研究第一部分計算 2000 年至 2019 年 65 歲以上老年人之住院感染事件發生率（每 1,000 人），由 2000 年的 68.8 至 2011 年提高至 90.7（平均年變化率+2.2%）。2011 至 2016 年間為高原期，2016 年後則轉為下降趨勢，截至2019 年為 76.3（平均年變化率為-3.5%）。85 歲以上老年人至 2019 年仍未有顯著下降趨勢。男性感染發生率為女性之 1.1 倍，且感染發生率均隨年齡顯著增加，85 歲以上及 75 至 84 歲的感染發生率分別為 65 至 74 歲的 4.2 及 2.2 倍。以泌尿道感染為例，2000 年發生率分別為 31.7，2019 年則為 36.3，65 至 74 歲、75 至84 歲及 85 歲以上各年齡組平均發生率依序為 20.6、51.8、109.5。研究對象於初始住院感染事件的院內死亡率為 7.9%，56.3%於研究期間內發生重複感染事件，平均間隔 826.4 天。隨重複感染次數增加，院內死亡率、重複感染發生率、入住加護病房比例、住院天數、醫療花費均增加，間隔天數則縮短。至第三次重複感染事件之院內死亡率達 11.5%，68.3%發生重複感染事件、平均間隔縮短至 274.7 天。當第一次住院感染事件為局部感染時，後續重複感染事件有更高機會出現相同感染類型。如初始住院感染為膽囊炎者，第一次重複感染事件發生膽囊炎的比例為 42.5%，遠超過泌尿道感染(25.5%)、肺炎(24.1%)等發生比例。 本研究第二部分篩選 2011 年曾有住院感染事件的 65 歲以上老年人，排除未完整存活於軌跡觀察期及基本資料缺失者後納入 79,666 人。依群組化軌跡模型分析分為四組：Infrequent infection (73.6%)、Ascending frequency of infection (12.2%)、Descending frequency of infection (11.4%)、Frequent infection (2.8%)。各組別中以 Infrequent infection 最為年輕，多重共病衰弱指數亦最輕微。其餘各組之平均年齡與進入不良結果追蹤期前的共病情形由低到高依序為 Descending frequency of infection、Ascending frequency of infection 及 Frequent infection，且各組均與 Infrequent infection 存在顯著差異。性別方面各組均為女性較多，唯有Frequent infection 女性較少(47.3%)，與 Infrequent infection 達統計顯著差異。此外，比較研究對象進入研究世代前及進入結果追蹤期前的共病情形，顯示四組不同長期感染軌跡組別，不僅在前後的共病狀況均有差異，且多重共病衰弱指數的惡化速度亦有明顯區別，與研究對象於軌跡觀察期內的感染頻率呈正相關。5 年結果追蹤期內，經校正各組間的性別、年齡及共病症後，各組別發生全死因死亡之風險與感染軌跡組別具有相關性。以 Infrequent infection 做為對照組，Frequent infection 組的 HR 為 3.00 (95% CI: 2.85-3.16)、Ascending frequency of infection 為 2.18 (95% CI: 2.11-2.25)、Descending frequency of infection 則為 1.87 (95% CI: 1.81-1.93)。 [研究結論] 臺灣老年人之感染症發生率高原期落在 2011 年至 2016 年間，男性及高齡者之發生率較高。85 歲以上老年人的感染發生率至 2019 年仍維持在高原期，需要臨床端更密切留意。老年人感染症負擔隨著重複感染次數的增加而擴大，重複感染亦更容易產生與初始感染相同的感染類型，顯示臨床工作者應積極介入控制共病及危險因子，降低老年人再發生重複感染事件的風險與負擔。老年人發生感染事件之頻率軌跡變化可區分為四組，感染軌跡組別與全死因死亡具有顯著相關，且各組之感染頻率與全死因死亡風險具有劑量效應，顯示長期的感染發生情形較單一事件與不良事件的發生具有更大的相關性。無論單次感染的嚴重度，當老年人重複暴露感染時仍有較高的死亡風險，故致力於減少老年人再發生感染症的可能對改善長期預後具有相當重要的臨床價值。

Background: Infection is one of the most important causes of mortality in the elderly. It not only leads to short-term disease burdens but may be associated with long-term adverse outcomes. However, studies regarding nationwide incidence and disease burden of infection in Taiwan are very limited. Existing studies on the association between infections and long-term adverse outcomes are also limited to specific infections, such as pneumonia. However, repeated infections in one individual may pose a higher risk of adverse outcomes. Objectives: The aims of this study were 1) to estimate the nationwide incidence of infections and associated impacts on disease burden between repeated infections in the elderly in Taiwan; and 2) to identify distinct infection trajectories of longitudinal infection episodes and examine the association between risk of all-cause mortality among different infection trajectories. Methods: The first part of this study is a retrospective cross-sectional study using data collected from Taiwan’s National Health Insurance Research Database (NHIRD). Individuals aged 65 years and older from 2000 to 2019 with any in-hospital diagnosis code related to infections were retrieved for the estimation of the incidence rate of infection. Each person would be calculated only once for the same infection type in the same year, and incidence rates were stratified by age and gender. Joint regression was used to examine the changes in incidence trends of infection, and Poisson regression was used to compare the incidence rate ratio between groups. The clinical and economic burdens associated with repeated infections were also analyzed. The second part of this study is a retrospective cohort study. Older people aged 65 years and older with any in-hospital diagnosis code related to infections (urinary tract infection (UTI), pneumonia, sepsis, cellulitis, cholecystitis, peritonitis, endocarditis, and meningitis) in 2011 were identified from the NHIRD, with the first admission date as the index date. For individuals surviving through a five-year trajectory period after the index date, infection episodes were retrieved every three months, and the group-based trajectory model (GBTM) was used to identify distinct infection trajectories among them. Another five years after the trajectory period was set as the follow-up period for adverse clinical outcomes. Kaplan-Meier curve and Cox proportional hazard model were used to examine the association between infection trajectories and all-cause mortality. Results: The incidence rate of infection increased from 68.8 in 2000 to 90.7 in 2011 (per 1,000 persons) in the elderly in Taiwan. A plateau in the incidence rate of infection was found between 2011 (90.7) and 2016 (85.0), and the incidence rate decreased thereafter (from 2016 to 2019 (76.3)). The incidence of infection significantly decreased in those aged 65-74 and 75-84 after 2011 but remained high between 2010 and 2019 in those aged over 85. The incidence rate ratio of males versus females was 1.1, and the incidence rate in those aged over 85 and 75-84 were 4.2 and 2.2 times higher than those aged 65-74, respectively. Take UTI as an example, the incidence was 31.7 (per 1,000 persons) in 2000 and 36.3 in 2019, and the average incidence in those aged 65-74, 75-84, and over 85 were 20.6, 51.8, and 109.5, respectively. The in-hospital mortality rate was 7.9% in index events, with 56.3% of individuals happening to repeated events, and the gap between infections was 826.4 days on average. Compared to the index infection event, repeated infections confronted more in-hospital death, re-infection, ICU utilization, medical costs, and longer length of stay with shorter time gaps between infections. Repeated events were more likely to be the same type of infection when the index event was a localized infection. For those with cholecystitis as an index event, the first repeated event had 42.5% cholecystitis, far more than UTI (25.5%) or pneumonia (24.1%). For the second part of the study, 79,666 older people were identified and divided into four distinct trajectories: infrequent infection (73.6%), ascending frequency of infection (12.2%), descending frequency of infection (11.4%), and frequent infection (2.8%). After adjusting for age, gender and comorbidities, compared to infrequent infection, the hazard ratio (HR) of all-cause mortality for frequent infection, ascending frequency of infection, and descending frequency of infection were 3.00 (95% confidence interval (CI): 2.85-3.16), 2.18 (95% CI: 2.11-2.25), and 1.87 (95% CI: 1.81-1.93), respectively. Conclusions: This study provides nationwide epidemiology of infection and associated clinical and economic burden on the elderly in Taiwan. While the incidence of infections decreased between 2011 and 2016, those aged over 85 still had high incidence rates of infection. The clinical and economic burden increased with subsequent infection episodes, showing that healthcare providers should make efforts to control comorbidities and risk factors to reduce repeated infections. Four distinct infection trajectories were identified, and dose-dependent risk of all-cause mortality was observed in those with frequent infection. Our findings suggest that repeated infections have incurred significant mortality risk in the elderly.