New methods for estimating follow-up rates in cohort studies

Table 2 The follow-up rate at each annual interval after subjects (N = 610) in a retrospective cohort study of 3-year and 5-year prostate cancer (PrCa) recurrence risk based on electronic medical record (EMR) data

Follow-up^b	N^c	Percentage Method η _percentage	Estimated follow-up using the formal method η _FPT	Clark’s completeness index η _CCI	Simplified Person-time Method η _SPT
1 Year	558	91.4%	95.7%	95.5%	95.7%
2 Year	472	86.2%	95.0%	94.5%	95.0%
3 Year	383	80.9%	93.6%	92.9%	93.8%
4 Year	295	75.6%	92.5%	92.3%	93.3%
5 Year	197	67.5%	91.8%	91.8%	93.0%

Follow-up rates were estimated using four methods, namely, (i) the standard Percentage Method (Eq. 1 ^a), (ii) the formal Method (FPT, Eq. 4), (iii) the Clark’s completeness index (CCI, Eq. 2), and (iv) the Simplified Person-Time Method (SPT, Eq. 5)
^aEquations are shown in the next
^bA retrospective cohort study was conducted among incident PrCa patients who underwent robotic assisted laparoscopic prostatectomy (RALP) by a single surgeon at Montefiore Medical Center (MMC) in the Bronx from 10/2005 through 12/2012. These subjects were followed for disease recurrence or progression through December 2012. A total of N = 610 PrCa patients who underwent RALP and had their follow-up at MMC were included in this analysis
^cWe calculated the follow-up rate at each year among the subset of the patients who had RALP early enough to be eligible for such length of follow-up. For example, to estimate the three year follow-up rate, we calculated this rate among the subjects who had RALP at least before 12/31/2009

ISSN: 1471-2288