目的：回診追蹤是癌症防治重要的一環，惟目前探討病人中斷回診追蹤的文獻仍不多見且結果不一，有鑑於此，本研究擬結合人口屬性、不同癌別及疾病特徵檢測其對回診追蹤之影響力並瞭解中斷原因，供臨床照護者發展更完善之照護模式時之參考。方法：資料來源為南部某醫學中心之癌症登記資料庫，研究對象為5,623人，而中斷回診者有471人，以羅吉斯迴歸模式進行影響因素分析。另，隨機抽樣219位存活者進行結構式「中斷回診追蹤問卷」之電話訪談，有效回收176份。結果：就研究對象之個別屬性而言，女性、39歲以下、居住其它縣市、罹患婦癌／乳癌／內分泌及神經系統者、0期／1期患者以及確診後第4年者中斷回診追蹤之比例相對較高。而二元羅吉斯回歸之分析也發現，年齡、居住地、整併期別、癌別與診斷年分具顯著預測力，整體模式之解釋力及適配度呈現良好。中斷原因以轉院為主，其次是等候時間太長、距離太遠與自覺身體狀況良好；再者，不同年齡層之中斷回診原因有所不同。結論：面對罹癌人數逐年增加且逐漸年輕化的趨勢，如何提供年輕族群（39歲以下）、早期期別、不同癌別等特定族群有效的衛教活動，進而提升定期回診追蹤之意願，值得臨床照護提供者及衛生政策制訂者思考。

Objectives: For 35 years in a row, cancer has been the leading cause of death in Taiwan, and follow-up is considered an important measure for the prevention of recurrence. To date, however, there are few studies related to interrupted follow-up issues. Further, the results and conclusions are inconsistent. Therefore, this study aimed to contribute to existing literature by examining the predictive effect of patient's demographic and disease's characteristics on the causes of interrupted follow-up in a more comprehensive approach. Methods: Data were extracted from the Cancer Registry Database of a medical center in southern Taiwan. 5,623 cases were included. There were 471 cases with interrupted follow-up by July 14, 2017. Binary logistic regression statistics were analyzed to examine the predictive model performance. In addition, a telephone survey with structured questionnaire was implemented among randomly selected 219 survivors, with 176 valid responses. Results: Overall, female patients, aged under 39, residing outside the hospital's city, with gynecological cancer/neuroendocrine tumors, stage 0 or 1, or in the fourth year after diagnostic confirmation, were relatively more likely than their counterparts to stop their follow-ups. The binomial logistic regression model showed that age, residence, cancer stage and year of diagnosis were significant predictors. Transfer to other hospital was the main cause for interruption, followed by long waiting time at the physician's office, far-distance transportation from residence to hospital, and self-perception of good health. Besides, causes differed across different age groups. Conclusions: Facing increasing cancer prevalence and a trend of younger incidence age, health care providers need to pay more attention about enhancing the patient's willingness to conduct regular follow-up visit through effective design of health education program for each specific group of patients.