糖尿病用藥的隱形界線：SGLT2i 與 GLP-1 RA使用的社經與地區差異

Abstract

背景 糖尿病是全球重大公共衛生問題，到2050年全球糖尿病患者人數可能達到13.1億。超過一半患者面臨心血管疾病、慢性腎臟病風險，進而增加死亡率。新型藥物鈉-葡萄糖共轉運蛋白2抑制劑（SGLT2i）與類升糖素胜肽-1 受體促進劑（GLP-1 RA）已被證明可顯著降低糖尿病患者的心血管與腎臟併發症風險。國際治療指引均建議早期使用這兩類藥物，特別是在高風險患者中（例如合併心血管疾病或慢性腎臟病者）。然而，全球實際臨床應用率仍遠低於預期，且不同地區及族群間存在差異。台灣雖有健保給付政策但仍非列為第一線用藥，且對於不同社會經濟與城鄉族群之間的藥物可及性與使用情況缺乏系統性的探討。本研究旨在透過分析某醫學中心不同院區的病歷資料，探討SGLT2i與GLP-1 RA的使用情況與其社經與地區差異，以提供更全面的理解與實證依據。 目的 本研究利用該醫學中心北、中、南院區的病歷資料庫，探討病人使用SGLT2i與GLP-1 RA的社經與地區差異。 方法 本研究採用回溯性橫斷研究（retrospective cross-sectional study），分析2022年01月01日至2024年12月31日間該醫學中心內2型糖尿病（T2DM）病人的電子病歷資料。並排除病歷資料不完整或無法確認診斷的個案。本研究收集病人的人口學特徵（年齡、性別、BMI、居住地）、臨床共病（心血管疾病、心衰竭、慢性腎臟病、肥胖等）、以及藥物使用模式，以評估SGLT2i與GLP-1 RA的使用情況。 資料分析首先進行描述性統計，計算SGLT2i與GLP-1 RA的使用率，並比較不同地區及族群間的差異。再採用羅吉斯回歸分析（multivariable logistic regression），評估影響SGLT2i與GLP-1 RA使用的因素，包括人口學變數、共病情況、社會經濟因素。本研究進一步針對所有研究藥物的使用族群，進行次分析（subgroup analysis），分析病人是否自費使用藥物，再次評估各影響因子對SGLT2i與GLP-1 RA使用的影響。 本研究所有數據皆進行匿名化處理，並已獲得台大醫院倫理審查委員會（e-REC/ IRB）核准，案號：202504148RINA。 結果 本研究共納入 2022 年至 2024 年間 101,933 位第二型糖尿病病人，其中 30,789 人（30.2%）曾於追蹤期間使用 SGLT2i 或 GLP-1 RA，71,144 人（69.8%）未使用；整體樣本中，27.9% 曾使用健保給付研究藥物，1.8% 曾於門診自費使用研究藥物。 多變項羅吉斯回歸分析顯示，在「是否使用研究藥物」方面，社會經濟變項整體影響有限。以低收入組為參照，中收入組與高收入組之校正勝算比(aOR)分別為 1.00（95% CI 0.96–1.04）與 1.00（95% CI 0.95–1.05），均未達統計顯著；相較於大學以下學歷，大學及以上教育程度者之 aOR 為 1.04（95% CI 0.99–1.10），亦未達顯著。相較之下，較年輕、男性、BMI >24 kg/m²、HbA1c >8%、腎功能仍介於 eGFR G1–G3、合併蛋白尿、心血管疾病或慢性腎病者，較可能使用研究藥物；其中合併心血管疾病與慢性腎病者之 aOR 分別為 1.84（95% CI 1.76–1.92）與 1.69（95% CI 1.61–1.76）。此外，在不同院區之間，相較南區病人，北區病人與中區病人，其用藥勝算較低，aOR 分別為 0.45（95% CI 0.42–0.48）與 0.43（95% CI 0.41–0.46）。 次分析中，納入 30,293 位曾使用任一研究藥物且具門診批價紀錄之病人，其中 1,863 人（6.2%）曾於門診自費使用研究藥物，其餘 28,430 人（93.8%）為健保給付。描述性分析顯示，自費比例於北區最高（8.3%）、中區次之（6.1%）、南區最低（1.9%）；低、中、高收入組之自費比例分別為 4.2%、6.1% 與 8.0%，大學及以上者為 11.1%，高於未滿大學者之 5.6%。多變項分析顯示，高收入者（aOR 1.42，95% CI 1.23–1.65）、大學及以上教育者（aOR 1.49，95% CI 1.30–1.70）、穩定工作族群（aOR 1.24，95% CI 1.01–1.51）及北區就醫者（相較南區 aOR 3.77）之自費機會顯著較高。 結論 本研究顯示，在台灣全民健康保險制度下，第二型糖尿病病人是否使用 SGLT2i 或 GLP-1 RA，主要仍受臨床需求、疾病嚴重度及心腎風險影響，社會經濟因素對整體用藥與否之影響相對有限。但在已使用研究藥物者中，是否進一步選擇門診自費，則受到收入、教育程度、職業穩定性及地區差異影響。換言之，健保制度雖在一定程度上縮小了不同社經族群於「能否使用新型藥物」上的差距，但當病人需自費取得新型藥物時，社會經濟不平等與區域落差仍然存在。此結果顯示，台灣糖尿病新型藥物的可近性，呈現出「健保內相對公平、健保外仍有不均」的雙層結構，未來政策上應朝向更符合心腎保護證據之給付設計，並關注高風險但支付能力不足族群之治療可近性。由於本研究資料僅來自單一醫學中心，上述發現應視為具機構特異性之結果，尚無法直接推論全民健康保險制度本身之效果。

Background: Type 2 diabetes mellitus (T2DM) is a major public health challenge and is closely associated with cardiovascular disease and chronic kidney disease. Sodium-glucose cotransporter-2 inhibitors (SGLT2i) and glucagon-like peptide-1 receptor agonists (GLP-1 RA) have demonstrated substantial cardiorenal benefits, and international guidelines recommend their early use in high-risk patients. However, real-world uptake remains suboptimal, and disparities may exist across regions and socioeconomic groups. In Taiwan, although National Health Insurance (NHI) provides partial reimbursement for these agents, restrictive reimbursement criteria may still affect access. Objective: This study aimed to investigate the utilization of SGLT2i and GLP-1 RA and to examine socioeconomic and regional disparities in their use within a multi-campus academic medical center in Taiwan. Methods: This retrospective cohort study included patients with T2DM treated at a medical center with northern, central, and southern branches between January 1, 2022 and December 31, 2024. Demographic characteristics, socioeconomic indicators, clinical comorbidities, laboratory data, and medication use were extracted from the institutional electronic medical record database. Multivariable logistic regression was used to identify factors associated with the use of SGLT2i or GLP-1 RA. A subgroup analysis was further performed among medication users with outpatient billing records to examine factors associated with self-paid use versus NHI-reimbursed use. Results: A total of 101,933 patients with T2DM were included. Among them, 30,789 (30.2%) used SGLT2i or GLP-1 RA during the study period, while 71,144 (69.8%) did not. Overall, 27.9% had used reimbursed study medications and 1.8% had used them through outpatient self-pay. In multivariable analysis, socioeconomic variables were not significantly associated with overall medication use: compared with the low-income group, the adjusted odds ratios (aORs) for the middle- and high-income groups were both 1.00, and college education was also not significantly associated with medication use (aOR 1.04). Instead, medication use was more strongly associated with clinical factors, including younger age, male sex, BMI >24 kg/m², HbA1c >8%, preserved-to-moderately reduced renal function, albuminuria, CVD, and CKD. Patients with CVD and CKD were more likely to receive study medications, with aORs of 1.84 and 1.69, respectively. In the subgroup analysis of 30,293 medication users with outpatient billing records, 1,863 patients (6.2%) had self-paid use. Self-pay was more common in the northern region (8.3%) than in the central (6.1%) or southern region (1.9%). A clear socioeconomic gradient was observed: self-pay rates were 4.2%, 6.1%, and 8.0% in the low-, middle-, and high-income groups, respectively, and 11.1% among those with a college education or above. In multivariable analysis, higher income (aOR 1.42), college education or above (aOR 1.49), stable occupation (aOR 1.24), and receiving care in the northern region (vs. southern region) were significantly associated with greater odds of self-paid use. Conclusions: Under Taiwan’s NHI system, the overall use of SGLT2i and GLP-1 RA appears to be driven primarily by clinical need and cardiorenal risk rather than by socioeconomic status. However, among patients who used these medications, self-paid use was strongly associated with income, education, occupational stability, and region. These findings suggest a two-layer pattern of access: relative equity within the reimbursed system, but persistent socioeconomic and regional inequality when access depends on out-of-pocket payment. Policy reform should therefore consider broader, evidence-based reimbursement criteria and improved access for high-risk patients with limited financial capacity. As data were drawn from a single medical center, these findings should be interpreted as institution-specific and cannot directly establish the effect of the NHI system. Acknowledgements The authors would like to express thanks to the staff of National Taiwan University Hospital-Statistical Consulting Unit (NTUH-SCU) for statistical consultation and analyses and the staff of Department of Medical Research for providing clinical data from National Taiwan University Hospital-Integrative Medical Data Center (NTUH-iMD).