糖尿病族群新使用基礎型胰島素之低血糖風險與利用研究

Hypoglycemic risk and utilization in diabetic population initiating basal insulin treatment

10.6342/NTU.2009.00264

葉姿辰

胰島素 ； 低血糖 ； 醫療利用 ； 病例交叉設計 ； 傾向性評分 ； insulin ； hypoglycemia ； health care utilization ； case-crossover design ； propensity score

臺灣大學預防醫學研究所學位論文

2009年

碩士

賴美淑

繁體中文

背景:國外研究顯示長效胰島素類似物(long-acting insulin analogue)因為濃度變化穩定，較無傳統中長效胰島素(intermediate-long acting insulin, ILI)易發生低血糖副作用，病患的醫療費用也並不會較高。台灣近年上市新型胰島素類似物使用率增加，然而台灣上市後藥品效性比較之研究缺乏胰島素主題，也無研究探討使用不同基礎型胰島素的糖尿病患其醫療利用與費用。本研究利用健保資料庫以傾向性評分方法(propensity score method)探討影響選擇長效胰島素類似物處方之因素，分析使用不同基礎型胰島素病患之醫療利用與費用，並設計一病例交叉研究(case-crossover study)，評估不同基礎型胰島素之低血糖風險。 研究方法: 本研究以93年1月1日至95年12月31日之全民健保資料庫做回溯性，觀察性研究，取93年7月1日至95年12月31日新使用長效胰島素類似物或傳統中長效胰島素之糖尿病患(19075人)，先以過去背景資料(開第一張處方前六個月)做為傾向性評分模型之變項，計算病患接受長效胰島素類似物的機率。另取發生過低血糖病患做病例交叉設計之研究對象(930人)，第一次發生低血糖為事件日，取事件日前一病例期與三組對照期(1:3 配對比例)，在條件式邏輯斯迴歸(conditional logistic regression)模型放入與時間相關之變項，得到降血糖藥物之暴露勝算比。再者將族群挑選追蹤期間單一使用研究藥品病患(17000人)，類似物分為insulin glargine(IGlar)組2961人及insulin detemir(IDet)組683人，及傳統中長效胰島素ILI組13356人，以描述性分析與事後檢定三組醫療利用與費用之差異。 結果: 第一部分:傾向性評分結果，醫學中心、診所、內分泌科、家醫科及女醫師較易傾向開立長效胰島素類似物，病人罹患代謝疾病、接受較多次糖化血色素檢驗、曾接受預混型胰島素藥物、接受口服降血糖藥物種類越多、或糖尿病門診費用越高者，都有較高機會接受長效胰島素類似物。第二部分: 使用傳統中長效胰島素比未使用該藥者增加低血糖風險[Odds ratio (OR) = 1.45, p<0.01]，使用長效胰島素類似物比未使用該藥者低血糖風險雖增加但不顯著[OR=1.57, p=0.08]，長效胰島素類似物與傳統中長效胰島素相較之低血糖風險無差異[OR=1.05, p=0.85]。第三部分: 醫療利用分析，長效胰島素類似物組(insulin glargine及insulin detemir組)之門診降血糖藥費及門診醫療費用高、門診次數高、門診研究藥品使用量多、及檢驗糖化血色素次數較多(p<0.001)，然而住院次數及住院費用較低(p<0.001)，使得整體醫療費用較少(IDet台幣104,769元，IGlar 台幣103,111元，ILI 台幣140,413元, p<0.001)。 結論: 使用傳統中長效胰島素的確比未使用該藥會增加低血糖風險；由於使用長效胰島素類似物樣本數少，傳統中長效胰島素與長效胰島素類似物之低血糖風險無差異的結果須保守看待，需要將來收集更多樣本作驗證。傳統中長效胰島素使用者之整體醫療費用較長效胰島素類似物使用者高，主要是高額的住院費用所貢獻。就目前研究結果，使用傳統中長效胰島素要注意低血糖副作用，同時注意病患是否因共病程度多，以至於住院情況較嚴重影響醫療利用大幅上升，若能改善醫療照顧品質(例如增加檢測糖化血色素次數，積極控制糖尿病)進而減少併發症發生，以致住院率與住院費用下降，或許能降低整體醫療費用。

Background: The usage of insulin analogue in Taiwan increase in latest years. Recent studies stated that long-acting insulin analogues (LAIA) provide a delayed onset of action and a peakless effect that may be associated with fewer episodes of hypoglycemia than conventional insulin. It might also provide moderation of health utilization and medical cost. However, there is lack of studies in postmarketing comparative effectiveness or health utilization of basal insulins in Taiwan. The aims of this study were to evaluate hypoglycemic risk of different basal insulin and to analyze the utilization from patients newly initiated these agents Besides, we evaluated which factors would influence prescriptions of insulin analogue with propensity score (PS) method. Methods: We conducted a retrospective analysis using National Health Insurance claims data for diabetes newly treated with LAIA or conventional intermediate-long acting insulin (ILI) between July 1, 2004 and December 31, 2006 (n=19075). Eligibility required at least 6 months of insurance before the index prescription date. The PS for receiving LAIA was estimated using logistic regression based on observed characteristics such as patient levels, hospital levels, and diabetes-related costs and utilization. Patients who had at least one hypoglycemic event during this period were extracted (n=930). The first hypoglycemic event was defined as index event. We defined case period as 1-14 days immediately before the event and 3 matched control periods 29-42, 43-56, 57-70 days prior to the index event of the same patient. Exposure to LAIA or ILI during different periods was compared using conditional logistic regression adjusting with time-dependent covariates. n the cohort study, individuals with prescriptions for only insulin detemir (IDet, n=683), insulin glargine (IGlar, n=2961) or ILI(n=13356) were followed until death or the end of study. Utilization of these groups was analyzed with general linear model and post hoc analyses. Results: Part I. The associated factors related to prescribing LAIA were medical centers, clinics, endocrinologists, family physicians, female doctors, metabolic disease, A1c tests, premixed insulin, or diabetes-related outpatient costs. Part II. By using case-crossover approach, recent use of ILI was associated with hypoclycemia. [Odds ratio (OR) = 1.45, p<0.01]. There was no significantlly increasing hypoglycemic risk of LAIA [OR=1.57, p=0.08]. The OR of LAIA over ILI was 1.05 and non-significant (p=0.85). Part III. On average, IDet and IGlar subjects had obviously more outpatient visits and costs, analogues usage, diabetes-related pharmacy costs and A1c tests (p<0.001) but fewer inpatient visits and costs (p<0.001) compared to ILI users. Total medical costs were reduced in analogues users (IDet NT$104769, IGlar NT$103111 vs ILI NT$140413, p<0.001). Conclusion: Subjects newly treated with intermediate-long acting insulin have indeed higher hypoglycemic risk than nonusers of this drug. Owing to fewer subjects of long-acting insulin analogues, we couldn’t confirm that there was no statistical difference of hypoglycemic risk between long-acting insulin analogues and intermediate-long acting insulin. We need to collect more information to prove it in future study. Compared to intermediate-long acting insulin users, long-acting insulin analogues users incurred higher outpatient utilization and diabetes-related pharmacy costs but fewer hospitalizations which led lower total costs. Finally, we should improve quality of care, such as increasing A1c tests, in those patients using intermediate-long acting insulin who have more comorbidities and higher hospitalization rates, thus diabetic complications would be decreased and total medical cost would be reduced thereafter.

1. 蔡嘉勳(2017)。含麩醯胺酸及膳食纖維之管灌配方對糖尿病患者營養狀況之改善。中山醫學大學營養學研究所學位論文。2017。1-99。