- 學位論文
台灣腎絲球過濾率估算公式對慢性腎臟病人心血管疾病住院、透析及 死亡風險的預測力評估
To evaluate the prediction of cardiovascular diseases hospitalizations, incident dialysis, and mortality by different estimated glomerular filtration rate equations in patients with chronic kidney diseases
摘要
研究背景：慢性腎臟病已被公認為全球公共衛生的問題，然而慢性腎臟病分期主要透過蛋白尿與腎絲球過濾率來定義，目前台灣臨床常用的腎絲球過濾率估算公式為simplified MDRD 公式，此一公式主要從西方國家所發展，因此不一定適合台灣。Taiwanese MDRD 與Taiwan-intercept MDRD 公式能夠更為準確的預測真實的腎絲球過濾率，然而目前沒有研究評估不同公式對臨床不良預後的預測性。因此本研究藉由比較在慢性腎臟病人不同公式所估算的腎絲球過濾率對不良預後（心血管住院、透析及死亡）風險的預測力。 研究方法：本研究利用南部某一醫學中心自91 年11 月到97 年5 月加入慢性腎臟病整合型照護計畫的病患，使用simplified MDRD 公式、CKD-EPI 公式、Taiwanese MDRD、Taiwan-intercept MDRD 公式估算病患開始收案時的腎絲球過濾率，追蹤至97 年底，觀察期間因心血管住院、進入透析及發生死亡的風險。Cox proportion hazard 模式用來校正與評估相關的風險。以適合度、區辨能力與重新分類能力來評估不同公式所估算的腎絲球過濾率對風險的預測性。統計以SPSS 19.0 進行分析，p<0.05 視為統計上的顯著差異。 研究結果：總共有2,181 名慢性腎臟病納入研究，排除照護小於90天的454 人，共有1,727 人納入追蹤，期間共有78 人因心血管疾病住院、510 人進入透析、140人發生死亡。整體而言，不論是CKD-EPI 公式、Taiwanese MDRD 或Taiwan-intercept MDRD 公式所估算的腎絲球過濾率在透析風險的預測力上皆顯著的低於simplified MDRD 公式所估算的腎絲球過濾率的預測力。但CKD-EPI公式所估算的腎絲球過濾率在死亡風險模式的重新分類能力顯著優於於simplified MDRD 公式所估算的腎絲球過濾率。但在大於65 歲、女性及糖尿病的患者上發現Taiwanese MDRD、Taiwan-intercept MDRD 公式所估算的腎絲球過濾率在透析的風險預測力上顯著優於simplified MDRD 公式所估算的腎絲球過濾率。 結論：本研究證實不論用哪一公式所估算的腎絲球過濾率對於不良的預後都有顯著預測力。Taiwanese MDRD 與Taiwan-intercept MDRD 公式相較於simplified MDRD 公式， 在大於65 歲、女性及糖尿病族群有較佳透析風險的預測力。本研究的發現，值得進一步的大型研究來證實。
並列摘要
Background: Chronic kidney disease (CKD) has been considered as a global public health problem in the world. The definition of CKD is based on criteria of proteinuria and level of glomerular filtration rate (GFR). Simplified MDRD equation for estimating GFR is most commonly used in clinics. It is not guaranteed that applying the equation to Taiwanese would not result in bias for GFR. Therefore, Taiwanese MDRD equation and Taiwan-intercept equation had been proved more accurate estimation for GFR. However, the performance of predicting risk through estimated GFR (eGFR) by using these equations was not well evaluated. The aim of the study is to compare the performance of models to predict adverse outcomes (cardiovascular disease hospitalization, dialysis and mortality) in different eGFR that were calculated by various equations. Materials and methods: Patients, who had CKD and joined an integrated care program from Dec. 2002 through May. 2008 in one medical center, southern Taiwan, were included to this study. We used simplified MDRD equation, CKD-EPI equation, Taiwanese MDRD equation and Taiwan-intercept equation to estimate various GFR at baseline. We followed up occurrence of cardiovascular diseases hospitalization, dialysis, and mortality to 2008. Cox proportion hazard model was used to estimate the risk of adverse outcomes. We used the statistical value regarding goodness of fit, calibration, and reclassification of models to identify the performance of risk prediction. All statistical analyses were performed by SPSS 19.0 and p<0.005 is considered as statistically significant. Results: Totally, 2,181 patients were included into the study. After excluding the patients joining the integrated care program less than 3 months (n=454), 1,727 patients were followed up regularly. In general, eGFR that were calculated by CKD-EPI equation, Taiwanese MDRD equation and Taiwan-intercept equation respectively provide significant worse risk perdition for dialysis than eGFR that was calculated by simplified MDRD equation. On the contrast, eGFR that was calculated by CKD-EPI equation has significant better reclassification of risk perdition for mortality than simplified MDRD equation. Furthermore, in elderly, female, and diabetes mellitus groups, the eGFR that were calculated by Taiwanese MDRD equation and Taiwan-intercept equation had significant better performance of risk predictions for dialysis than simplified MDRD equation. Conclusion: Taiwanese MDRD equation and Taiwan-intercept equation provide a better risk predicting models for dialysis in old age, female, and diabetes mellitus groups. Our findings need more large-scale studies to further confirm.