成人族群身體質量指數與特定社區型感染症及感染症相關長期死亡率相關性:系統性回顧與統合分析

Abstract

背景 社區型感染症造成全球廣泛的發病率和死亡率，並且反映在住院人數的增加，為醫療體系帶來巨大的經濟負擔。長期以來體重不足一直被認為是營養不良的代表，也是引發學齡前兒童與老年人族群中社區型感染(community-acquired infections, 簡稱CAIs)的危險因子。肥胖則被認為與2009年全球H1N1 A型流感相關的住院率跟死亡相關，然而身體質量指數(body mass index,簡稱BMI)與常見CAIs的相關性尚未有清楚的定論。本研究將對BMI與社區型肺炎(community-acquired pneumonia,簡稱CAP)、泌尿道感染(urinary tract infections, 簡稱UTIs)、腹腔內感染(intra-abdominal infections,簡稱IAIs)、活動型結核菌(active tuberculosis,簡稱active TB)、真菌感染以及感染症相關長期死亡率之相關性進行系統性回顧與統合分析。 方法 我們使用關鍵字在三大文獻資料庫PubMed, Embase和Cochrane Library搜尋至2019年7月31日前發表的研究，並且只將世代研究納入分析。統合分析的過程將BMI轉換成連續變項，使用劑量效應分析觀察BMI與常見CAIs在線性模型下的趨勢，並採用隨機效應模型合併單一研究的相對風險值。在非線性模型中，我們使用restricted cubic spline模型分析兩者之間的關係。I2用來評估研究間的異質性，Egger’s test 用來評估是否有出版偏差，並使用次群組分析和meta-regression 分析試圖找出潛在異質性的來源。利用Newcastle-Ottawa quality assessment來評估觀察性研究的偏差風險。 結果 起初搜尋到5920篇文獻，最終有七篇討論BMI與CAP(n=2885663)，一篇討論UTIs(n=230410)，七篇探討IAIs(n=1547312)，沒有單純探討真菌感染相關文獻，二篇探討多重部位感染，六篇active TB(n=597666)，四篇感染症相關長期死亡率(n=276153)，總共收錄27篇世代研究。本研究結果發現統合的相對風險顯示BMI與CAP呈現J型曲線，轉折點落在BMI 25-30 kg/m2; BMI與UTIs風險呈現線性正相關，每增加五單位的BMI可增加5%的UTI風險(RR=1.05, 95%CI 1.00-1.10, I2=84%); 每增加五單位BMI所得到IAIs的相對風險為1.29(95%CI 1.19-1.39, I2= 83%)，為線性正相關; 每增加五單位的BMI得到活動型肺結核的統合相對風險為0.62(95%CI 0.44-0.89, I2=99%)，呈現線性負向關係；本研究也看到BMI與感染相關長期死亡率之風險呈現負向關係，觀察到肥胖矛盾的現象。從我們的研究中發現身高與體重取得方式可能為BMI與腹腔內感染以及活動性肺結核高度異質性的來源，卻沒有找到潛在BMI與感染症的修飾因子。 結論 這是第一篇以成人族群為對象探討BMI與常見社區型感染症相關性的統合分析。從我們的結果顯示，除了擬定對抗肥胖的策略外，成人體重不足的相關議題值得關注並加以預防。然而在肺結核感染的成人高風險群中，其健康的身高體重範圍定義相較其他常見社區型感染症可能有所不同。

Background Community-acquired Infections (CAIs) cause widespread morbidity and mortality which reflected increased hospital admission with a significant growing economic burden. Underweight has long been a proxy of undernutrition and was possible a risk factor of CAIs in preschool and the elderly, while obesity has just known a risk factor after 2009 pandemic influenza A. The effect body mass index (BMI) on common CAIs in adults is unclear. We conducted a systematic review and meta-analysis of the risk of community-acquired pneumonia (CAP), urinary tract infections (UTIs), intra-abdominal infections (IAIs), active tuberculosis (TB), fungal infection, and infection-related long-term mortality for BMIs. Materials and Methods Published studies were searched form PubMed, Embase and Cochrane Library through July 31, 2019 using key words. We only included cohort studies and used dose-response analysis to observe the linear associations between BMI and the risk of CAIs from individual study after converting BMI to continuous data. Random-effects analysis was applied to pool the effect size of each study. We also used restricted cubic spline (RCS) model in our meta-analysis as our non-linear model assumption to fit the relationship of BMI and CAIs. I2 was used to evaluate heterogeneity, and Egger’s test was performed to evaluate the publication bias. Subgroup analysis and meta-regression were also applied to quantify the potential source of heterogeneity between studies. Risk of bias was assessed using Newcastle-Ottawa quality assessment scale for cohort studies. Results A total of 5920 articles were in the initial search, and there were total 27 cohort studies met our inclusion criteria. Two cohort studies for BMI and multiple body parts for infection risk provided data for CAP, UTIs, IAIs and fungal infections. Nine cohort studies assessing the relationship between BMI and CAP risk (n=2885663), three cohort study for BMI and UTIs risk (n=230410), nine cohort studies for BMI and IAIs risk (n=1547312), one cohort study for BMI and fungal infections (n=37808), six cohort studies for BMI and active tuberculosis (n=549729), and four cohort studies for BMI and infections-related long-term mortality (n=276153). The pooled relative risk (RR) and RCS model revealed a J-shaped relationship between BMI and CAP with a change point falling on BMI 25-30 kg/m2. The positive linear association was noted between BMI and the risk of UTIs with 5% increase in the RR of UTIs for each 5 units increase in BMI (RR= 1.05, 95%CI 1.00-1.10, I2=84%). The summary RR of IAIs for a 5-unit BMI increment was 1.29 (95%CI 1.19-1.39, I2=84%) which revealed linear positive association between BMI and IAIs risk. The pooled RR was 0.54 (95%CI 0.47-0.61, I2=92%) if active TB for a 5-unit increase in BMI showing an inverse linear relationship between BMI and the risk of active TB. We also found that there was an inverse non-linear relationship between BMI and infection-related long-term mortality risk (RR=0.62, 95%CI 0.44-0.89, I2=99%). Assessment of weight and height may explain high heterogeneity between studies of BMI and IAIs, or active TB, but we did not find definite effect modifier between BMI and infections discussed in our study. Conclusions This is the first meta-analysis to define the association between BMI and common CAIs in adults. These results suggest that not only do we have to make strategies for combating obesity but also preventing underweight in adults. However, the healthy weight-for height bands could be defined differently in risky TB infected population.