OBJECTIVE: This study aimed to investigate transitions of recurrence hazard and peak recurrence time in patients with nonmetastatic CRC using the hazard function. SUMMARY OF BACKGROUND DATA: A postoperative surveillance period of 5 years is consistent across major guidelines for patients with nonmetastatic CRC, but surveillance intervals differ. Estimates of instantaneous conditional recurrence rate can help set appropriate intervals. METHODS: The study population consisted of 4330 patients with stage I to III CRC who underwent curative resection at the National Cancer Center Hospital between January 2000 and December 2013. Hazard rates of recurrence were calculated using the hazard function. RESULTS: Recurrence rates in patients with stage I, II, and III CRC were 4% (50/1432), 11% (136/1231), and 25% (424/1667), respectively. The hazard curve for stage I was relatively flat and hazard rates were consistently low (<0.0015) for 5 years after surgery. The hazard curve for stage II had a peak hazard rate of 0.0046 at 13.7 months, after which the curve had a long hem to the right. The hazard curve for stage III had an earlier and higher peak than that of stage II (0.0105 at 11.6 months), with a long hem to the right. CONCLUSIONS: Changes in recurrence hazard for CRC patients varied considerably by stage. Our findings suggest that short-interval surveillance might be unnecessary for stage I patients for the first 3 years after surgery, whereas short-interval surveillance for the first 3 years should be considered for stage III patients.
OBJECTIVE: This study aimed to investigate transitions of recurrence hazard and peak recurrence time in patients with nonmetastatic CRC using the hazard function. SUMMARY OF BACKGROUND DATA: A postoperative surveillance period of 5 years is consistent across major guidelines for patients with nonmetastatic CRC, but surveillance intervals differ. Estimates of instantaneous conditional recurrence rate can help set appropriate intervals. METHODS: The study population consisted of 4330 patients with stage I to III CRC who underwent curative resection at the National Cancer Center Hospital between January 2000 and December 2013. Hazard rates of recurrence were calculated using the hazard function. RESULTS: Recurrence rates in patients with stage I, II, and III CRC were 4% (50/1432), 11% (136/1231), and 25% (424/1667), respectively. The hazard curve for stage I was relatively flat and hazard rates were consistently low (<0.0015) for 5 years after surgery. The hazard curve for stage II had a peak hazard rate of 0.0046 at 13.7 months, after which the curve had a long hem to the right. The hazard curve for stage III had an earlier and higher peak than that of stage II (0.0105 at 11.6 months), with a long hem to the right. CONCLUSIONS: Changes in recurrence hazard for CRC patients varied considerably by stage. Our findings suggest that short-interval surveillance might be unnecessary for stage I patients for the first 3 years after surgery, whereas short-interval surveillance for the first 3 years should be considered for stage III patients.