目的：在 2011 年 1 月 1 日前，全民健康保險資料，並未記錄手術或放射治療之執行日期，因而 使正確的治療間隔時距之計算受到限制。我們的目的是推導並驗證手術與輔助性放射治療之間 隔時距的替代估計方法。 材料與方法：本研究自全民健康保險資料庫中提取並回溯分析於 2011 年 1 月 1 日至 2012 年 12 月 31 日期間被診斷為惡性腦瘤並接受手術與輔助性放射治療之個案。手術與輔助性放射治療間 隔時距之替代估計值，係依據門診就醫日期及住院首日計算得到。利用最佳配適線性迴歸，以 手術與放射治療執行日期計算之正確間隔時距為基準，以驗證該替代估計值。 結果：根據 ICD-9 診斷碼，1,527 例個案被定義為惡性腦瘤診斷。其中 394 例個案因存在其他 癌症診斷而被排除。在剩餘的 1,133 例個案中，463 例接受腦瘤切除手術併術後輔助放射治療 的個案被分析。手術與輔助放射治療間隔時距之替代估計值與正確值間的相關性佳，斜率為 1.0076，R2 等於 0.8633 且y 截距接近-1。藉由排除間隔時距小於或等於 0 之個案，本估計方法 可進一步最佳化至 R2 等於 0.9809 而斜率為 0.9951 且y 截距接近 5。 結論：我們提出了手術與輔助放射治療間隔時距之替代估計方法並驗證之。此研究提供一可能 方法，用以克服 2011 年 1 月 1 日前全民健康保險資料日期資訊不充分所造成的限制，使得在進 行手術與輔助放射治療間隔時距對癌症預後影響的相關研究時，可以充分利用衛生福利資料庫 中大多數的全民健康保險資料進行較大規模的研究。

Purpose : The date of surgery or radiotherapy (RT) was not recorded in the National Health Insurance database until Jan. 1, 2011, which limits the calculation of the exact time interval between treatments. Our purpose is to formulate and validate a surrogate estimate for the time interval between surgery and adjuvant RT. Materials and Methods : Information of patients with malignant brain tumors diagnosed and treated with surgery and adjuvant RT between Jan. 1, 2011 and Dec. 31, 2012 was obtained from the National Health Insurance (NHI) database and retrospectively analyzed. The surrogate estimate for the interval between surgery and RT was formulated according to the date of outpatient visits and the first date of hospitalization. Surrogate estimate of the interval was validated with the exact interval calculated by the later coded dates of surgery and RT from the database, using the best-fitting linear regression. Results : 1,527 cases of malignant brain tumors were defined according to ICD-9 diagnosis code. 394 cases were excluded for the presence of other cancer diagnoses. Among the remaining 1,133 cases, 463 cases receiving both brain tumor excision and adjuvant RT were analyzed. Surrogate estimate of time intervals and exact time intervals correlate well with a slope of 1.0076, a y-intercept close to -1 day, and an R2 of 0.8633. The calculation is further improved to an R2 of 0.9809 with a slope of 0.9951 and a y-intercept close to 5 days, by excluding cases with surrogate estimates of an interval smaller or equal to 0. Conclusion : Our validated surrogate estimate of the interval between surgery and adjuvant RT provides a possible method to overcome the limitation of inadequate date information before Jan. 1, 2011, and therefore enables the application of most of the data from the Health and Welfare Database for research regarding the effect of RT delay on cancer outcomes.