兒童早期益生菌攝取與性早熟之關聯：使用臺灣出生世代資料庫之回溯性世代研究

Abstract

研究背景 在臺灣，益生菌補充劑被廣泛使用，特別是在兒童族群當中。近期研究指出，腸道菌叢可能影響青春期發育時間。此外，發育時間異常已被證實與心絞痛、高血壓及第二型糖尿病等不良健康風險上升相關。雖然動物研究顯示益生菌可能延緩性早熟，但在人類族群中仍缺乏充分證據。因此，本研究旨在探討兒童早期攝取益生菌與青春期發育時間之關聯。 研究方法 本回溯性世代研究使用臺灣出生世代研究（Taiwan Birth Cohort Study, TBCS）資料進行分析；TBCS從2005年開始納入24,200對母嬰配對組資料，為一項具有全國代表性的縱貫性研究。本研究將TBCS中年齡介於6個月大至9歲之間、且同意將其資料串聯至全民健康保險研究資料庫（National Health Insurance Research Database, NHIRD）之兒童納入研究，並排除5歲以前被診斷性早熟的兒童。本研究以兒童早期益生菌攝取作為主要暴露，並透過問卷資料進行評估。性早熟結果則透過串聯NHIRD後，利用病歷資料中的ICD-9-CM以及ICD-10-CM診斷碼判定。本研究使用多變項Cox比例風險模型（Cox proportional hazards model）校正共變項後評估益生菌攝取與性早熟發生風險之間的關聯，並進行多項敏感度分析，包含性別分層模型、邏輯斯回歸模型（Logistic regression model）、校正身體質量指數（Body Mass Index model, BMI）之模型、工具變數分析（Instrumental variable analysis）以及負對照結果（Negative control outcome）。 研究結果 本研究共納入10,650名兒童，其中274名為性早熟病例。8,930名兒童在5歲前曾攝取過益生菌補充劑，包括4,695名男孩和4,235名女孩。另有1,720名兒童在5歲前未使用過益生菌補充劑，其中包含890名男孩以及830名女孩。在主要的Cox比例風險模型當中，校正後模型結果顯示在5歲前攝取益生菌補充劑的兒童相比從未攝取者，其性早熟風險高出89%（aHR: 1.89; 95% CI: 1.02–3.52）。在性別分層分析中，女孩的益生菌攝取與其性早熟風險上升97%有關（aHR: 1.97; 95% CI: 1.03–3.77）。校正BMI之模型結果亦與邏輯斯回歸模型之結果則顯示此關聯減弱。負對照結果中發現益生菌攝取與中耳炎之間出現無關聯（aHR: 0.87; 95% CI: 0.67–1.14），而以異位性皮膚炎作為工具變數的分析則顯示，經工具變數校正後的益生菌攝取與性早熟之間並無關聯（p=0.214）。 結論 在校正共變項後，我們發現兒童早期益生菌攝取與較高的性早熟發生風險有關，尤其在女孩當中。此關聯在邏輯斯回歸模型與校正BMI之模型當中並不明顯。負對照結果分析驗證了主要結果與模型的穩健性，而工具變數分析則顯示其存在未測量之干擾因子。然而，本研究受限於無法測量環境賀爾蒙的影響，未來仍需進一步研究以釐清其潛在機制。

Background Probiotic supplements are widely consumed in Taiwan, especially among children. Emerging evidence suggests that the gut microbiota might influence pubertal timing, and abnormal timing has been associated with adverse health outcomes such as heightened risks of angina, hypertension, and type 2 diabetes. Although animal studies suggest that probiotics may delay early puberty, there is a lack of conclusive evidence in human populations. This study aims to investigate the association between early childhood probiotic intake and pubertal timing. Materials and methods This retrospective cohort study analysed data from the Taiwan Birth Cohort Study (TBCS), a nationally representative longitudinal study enrolling 24,200 mother-infant pairs in 2005. We included children aged 6 months to 9 years who consented to the linkage of TBCS data with the National Health Insurance Research Database (NHIRD) and excluded children who were diagnosed with early puberty before age 5. We assessed early childhood probiotic intake as our primary exposure, using structured questionnaires to collect data on exposure. Early puberty outcomes were identified using ICD-9 and ICD-10 codes from a linked dataset of the TBCS and the NHIRD. We applied a multivariable Cox proportional hazard model adjusting for covariates, to evaluate the hazard of puberty associated with probiotic intake and conducted multiple sensitivity analyses, including sex-stratified models, logistic regression models, a BMI-adjusted model, an instrumental variable analysis, and a negative control outcome. Result A total of 10,650 children were included in this study, identifying 274 cases of early puberty. Regarding early childhood probiotic supplementation, 8,930 children were ever-users, consisting of 4,695 boys and 4,235 girls, while 1,720 were never-users, including 890 boys and 830 girls. We found that children who had consumed probiotic supplements in their early life had an 89% higher hazard of early puberty compared to never-users in the primary Cox proportional hazards model (adjusted HR: 1.89; 95% CI: 1.02–3.52). In the sex-stratified analysis, probiotic intake was associated with a 97% increase in the risk of early puberty among girls (adjusted HR: 1.97; 95% CI: 1.03–3.77). BMI-adjusted model and the logistic regression model results indicated that the association was attenuated. A negative control outcome indicated no association between probiotic intake and otitis media (adjusted HR: 0.87; 95% CI: 0.67–1.14), and an instrumental variable analysis using atopic dermatitis as the instrument showed no association between instrumented probiotic intake and early puberty (p=0.214). Conclusion After adjustment for covariates, we found that early childhood probiotic intake is correlated with a higher hazard of early puberty, especially among girls. This association was less evident in the logistic regression model and the BMI-adjusted model. The negative control analysis validated the robustness of our primary findings, while the instrumental variable analysis revealed the presence of unmeasured confounding. Further studies are needed to clarify the underlying mechanisms.