透析患者髖關節骨折與死亡之競爭風險分析

Competing risk analysis of hip fracture and death among dialysis patients

黃芬緯(Fen-Wei Huang)

指導教授 ： 梁文敏

共同指導教授 ： 林建甫(Chien-Fu Lin)

中國醫藥大學/管理學院/生物統計研究所碩士班/碩士(2014年)

摘要

背景：
　　在台灣透析已是一大重要議題，根據 2013 年美國腎臟病登錄系統 (USRDS) 所發表的統計資料顯示，調查全球四十二個國家中，台灣的透析發生率為每百萬人中有361人，為亞洲國家之首，超越日本的295人 (每百萬人)、新加坡的279人 (每百萬人)，名列第三。
　　病人在接受透析治療過程會造成骨病變。腎骨病變 (Osteodystrophy) 為透析患者主要合併症之一，有研究表示，若比較血液透析病人與健康族群骨密度的差異，結果發現血液透析病人的骨密度較健康族群低，而與一般健康族群相比較，透析患者髖關節骨折的發生率遠高於一般健康族群。
　　過去的研究顯示透析患者容易發生死亡，即便如此，很少有研究考慮死亡作為髖關節骨折的競爭事件，故本研究考慮將死亡作為髖關節骨折之競爭事件，探討髖關節骨折之發生風險及相關影響因子。
材料與方法：
　　資料來自衛生福利部中央健康保險署之「健康資料加值應用協作中心」的「 2000 年兩百萬歸人檔」，挑選自 2001 至 2007 年間，首次接受透析患者並追蹤至 2009 年。估計透析患者髖關節骨折之累積發生率、死亡之累積發生率，並以 Cox 等比例風險迴歸模型，分析年齡、性別、透析原因、透析方式以及共病症，對於髖關節骨折發生之影響，且更進一步將死亡作為髖關節骨折的競爭風險，以 Fine 和 Gray 迴歸模型分析其相關之危險因子。
結果：
　　透析髖關節骨折累積發生率在第1、2、5年分別是0.98%、1.81%、3.87%；透析死亡累積發生率在第1、2、5年分別是14.12%、25.39%、46.59%。
　　透析發生死亡之影響因子有年齡、男性、血液透析、透析原因為高血壓性腎衰竭、共病症個數≧1、糖尿病、充血性心肌衰竭、腦血管疾病、移轉性腫瘤、肝病、慢性併發症糖尿病、惡性腫瘤、半身或下半身麻痺。
　　在單變量部分，年齡、透析原因為慢性腎衰竭和共病症個數 ≥ 2皆為髖關節骨折的危險因子，性別也是屬危險因子但沒有統計上的顯著。共病症部分，糖尿病、腦血管疾病、消化道潰瘍、週圍心血管疾病是危險因子。多變量部分，年齡、性別、透析方式為血液透析、透析原因為慢性腎衰竭皆是髖關節骨折的危險因子，共病症個數≥ 2也是屬危險因子但沒有統計上的顯著，而中度或重度腎病、糖尿病也是危險因子但沒有統計上的顯著。
　　相較於未考慮死亡為競爭風險的結果，競爭風險分析結果顯示髖關節骨折累積發生率明顯下降許多。女性的髖關節骨折第5年累積發生率由3.91% 下降至1.1%，男性的髖關節骨折第5年累積發生率由3.85%下降至1.3%。這代表因為透析患者的死亡人數多，是發生髖關節骨折的12.4倍左右，故若未考慮死亡為競爭風險，髖關節骨折發生率會明顯高估許多。
結論：
　　考慮競爭風險下，髖關節骨折累積發生率明顯下降許多，表示若不考慮死亡之事件的話，髖關節骨折的發生則會出現高估情形。
　　年齡的增加、性別為男性、血液透析、透析原因為慢性腎衰竭、糖尿病、高血壓皆可能為髖關節骨折的危險因子。

關鍵字

全民健康保險資料庫透析 ； 髖關節骨折 ； Fine 和 Gray 迴歸模型 ； 競爭風險

並列摘要

Background:
Dialysis in Taiwan is one of the important issues. To the best of our knowledge, compared with the general population, the incidence of hip fractures of dialysis patients is higher than the general populations. Previous studies have shown that dialysis patients were prone to death, even so, few studies have considered death as hip fractures competitive events. In this study, we will consider death as a competitive event of hip fractures and discuss the risks of hip fractures and related factor.
Methods:
Subjects were the patients with first time dialysis selected from the National Health Insurance Research database (NHIRD) between 2001 and 2007. We estimated cumulative incidence of hip fractures in dialysis patients, and the cumulative incidence of death in dialysis patients. Cox proportional hazards regression model was used to analyze age, gender, dialysis reasons, dialysis modality and comorbidities. We consider what affect the occurrence of hip fracture with treating death as a competing risk of hip fractures. Fine and Gray regression model was used to analyze the associated risk factor and estimate the incidence of hip fracture.
Results:
Dialysis cumulative incidence of hip fractures in the 1,2,5 years were 0.98%, 1.81%, 3.87%, and dialysis cumulative incidence of death in the 1,2,5 years were 14.12%, 25.39%, 46.59%.
The risk factors of death occurred dialysis are elderly, male, hemodialysis, dialysis renal failure due to hypertension, larger number of comorbidities, diabetes, congestive heart failure, cerebrovascular disease, metastatic cancer, liver disease, chronic concurrent disease of diabetes, cancer, and Hemiplegia.
In univariate analysis, elerly, dialysis due to chronic renal failure, and larger number of comorbidities are risk factors for hip fracture. Gender is also a risk factor but not statistically significant. In comorbidities, diabetes, gastrointestinal ulcers, cerebrovascular diseases, and peripheral cardiovascular disease are risk factors. In multivariate analysis, elderly, male, HD, and dialysis due to chronic renal failure are risk factors.
From results of competition risk analysis, we found the cumulative incidence of hip fractures decreased a lot. The 5-year cumulative incidence rate in female hip fracture decreased from 3.91% to 1.1%, and the 5-year cumulative incidence rate in male hip fracture decreased from 3.85% to 1.3%. The reason of overestimation was due to a large proportion of death occurred compared with the occurrence of hip fracture with a ratio of 12.4 times in dialysis patients.
Conclusions:
Consider the competing risk of death, the cumulative incidence of hip fractures decreased significantly, which means that without considering the event of death, the occurrence of hip fractures would be overestimated. Increasing age, male gender, hemodialysis, dialysis due to chronic renal failure, diabetes, high blood pressure are likely to increase the risk of hip fracture.

並列關鍵字

National Health Insurance Research Database ； Dialysis ； Hip Fracture ； Fine and Gray regression model ； Competing Risk

參考文獻

4. 郭俐蘭, 蔡維河, 余錦美, 洪瑞伶, 楊芝青. 血液透析病人骨質疏鬆之危險因子. 北市醫學雜誌 2012;9:234-43.

6. 陳彥旭;林志慶;楊五常. 腎性骨病變. 腎臟與透析 2008;20:181.

13. 鄭振廷, 侯宏彬, 錢慶文. 影響洗腎病患定期血液透析醫療資源耗用之因素. 醫務管理期刊 2005;6:291-308.

16. 羅大維, 郭亮君, 唐憶淨. 髖關節骨折患者入住中期照護之成效探討. 台灣老年醫學暨老年學雜誌 2012;7:20-36.

32. 蔡敦仁. 慢性腎衰竭. 當代醫學 1994:604-15.