目的: 为中危非肌层浸润性膀胱癌确定最佳的膀胱镜监测频率方法: 回顾性分析2001年1月至2019年10月于北京大学人民医院行经尿道膀胱肿瘤电切术的中危非肌层浸润性膀胱癌患者，收集患者的临床、病理学和随访资料。术后2年每3至6个月对患者进行膀胱镜检测，根据患者复发和进展的情况，在术后2年假设3种强度的监测策略：模式1：间隔3个月，模式2：间隔6个月，模式3：间隔12个月，比较3种模式间检测复发和进展的延迟个数和时间差异。结果: 共纳入185例患者，其中男性144例(77.8%)，女性41例(22.2%)，中位年龄68 (59~76)岁。单发肿瘤有118例(63.8%)，多发肿瘤有67例(36.2%)。术后病理Ta期179例(96.8%)，T1期有6例(3.2%)。高级别病变108例(58.4%)，低级别病变77例(41.6%)。在术后2年随访期间内，52例(28.1%)患者出现复发，133例(71.9%)未出现复发；11例(5.9%)患者发生进展，174例(94.1%)未发生进展。相比于模式1，模式2出现复发检测延迟的个数为29例(55.8%)，模式3为41例(78.8%)，两者之间差异有统计学意义(P=0.012)。模式1、模式2和模式3检测复发的中位延迟时间分别为1.00、1.99和4.19个月。模式1与模式2和模式3之间差异均有统计学意义(P=0.001, P=0.013)。与模式1相比，模式2出现进展检测延迟的个数为5例(45.4%)，模式3为8例(72.7%)，两者之间差异无统计学意义。模式1、模式2和模式3检测进展的中位延迟时间分别为1.00、2.00和3.00个月，三者之间差异无统计学意义。结论: 6个月间隔的膀胱镜检测模式与3个月相比，虽然会稍微延迟检测到肿瘤复发和进展，但不会造成严重的不良结局，并能够减少患者的痛苦和经济负担，在中危非肌层浸润性膀胱癌是可行的。

Objective: To determine the optimal cystoscopic frequency for intermediate-risk non-muscle invasive bladder cancer. Methods: Patients with intermediate-risk non-muscle invasive bladder cancer, who underwent transurethral resection of bladder tumor in Peking University People's Hospital from January 2001 to October 2019, were retrospectively analyzed. Their clinical, pathological and follow-up data were collected. In postoperative 2-year period, the patients were underwent cystoscopy every 3 to 6 months. Depending on recurrence and progression of the patients, we hypothesized three strategies of surveillance intensity in the first 2 years after surgery: model 1: 3-month intervals, model 2: 6-month intervals, and model 3: 12-month intervals. The differences in the numbers and time of delayed detection of recurrence and progression were compared among the three models. Results: A total of 185 patients were enrolled, including 144 males (77.8%) and 41 females (22.2%). The median age was 68 (59-76) years. There were 118 cases (63.8%) with single tumor and 67 cases (36.2%) with multiple tumor. Of the patients 179 (96.8%) had stage Ta and 6 (3.2%) had stage T1. There were 108 cases (58.4%) with high-grade disease and 77 cases (41.6%) with low-grade disease. During the follow-up period of the first 2 years, 52 patients (28.1%) had recurrence, 133 cases (71.9%) had no recurrence, 11 cases (5.9%) had progression and 174 cases (94.1%) had no progression. Compared with model 1, 29 (55.8%) delayed detection of recurrence in model 2 vs. 41 (78.8%) delayed detection of recurrence in model 3, and the difference was statistically significant (P=0.012). The median delayed time of detecting recurrence was 1.00 months in model 1, 1.99 months in model 2 and 4.19 months in model 3, respectively. There were statistically significant differences between mode 1 and model 3 (P=0.001), and between model 2 and model 3 (P=0.013). Compared with model 1, 5 (45.4%) delayed detection of progression in model 2 vs. 8 (72.7%) delayed detection of progression in model 3, and the difference was not statistically significant. The median delayed time of detecting progression was 1.00 month in model 1, 2.00 months in model 2 and 3.00 months in model 3, respectively. There was no statistically significant difference among them. Conclusion: Although providing slightly slower detection of tumor recurrence and progression, compared with 3-month intervals of cystoscopy, 6-month intervals do not result in serious adverse outcomes and reduce cost and pain of the patients, which is feasible in intermediate-risk non-muscle invasive bladder cancer.