Study
PLCO
(Learn more about this study)
Project ID
2010-0273
Initial CDAS Request Approval
Sep 27, 2010
Title
Non-Compliance Adjustment Applying Instrumental Variable Approach (CES) to Prostate Mortality Data
Summary
The recent report from the PLCO Screening Trial by Andriole, et al (2009) provides the latest estimates of the effects of randomization to prostate cancer (PC) screening on mortality, with complete follow-up through 7 years, and 67% follow-up through 10 years. The relative mortality (and 95% confidence interval) from PC through 7 years is estimated at 1.13 (0.75, 1.70). The PLCO project plans another analysis upon the completion of 13 years of follow-up, as originally proposed. The currently available precision is low; a 25% reduction in PC mortality from screening cannot be ruled out. However, the 10 year data are consistent with this, and more precise, with relative mortality at 1.11 (0.83, 1.50). These results suggest the possibility that prostate cancer screening is in fact detrimental. Since PC screening was neither complete in the screen group, nor absent in the control group, then the actual effect of PC screening may differ substantially from the randomized effect. We propose to estimate this actual effect. During the first 6 years following randomization, Andriole, et al (2009) estimated the proportion of control group subjects who received recent PSA screening as ranging between 40% and 52%. The proportion of subjects who obtained at least one PSA test at any time during study years would be even higher. This latter proportion can also be estimated from study data. Approximately 15% of the control group can be individually classified in terms of whether they received either PSA or DRE testing. For the screen group, adherence to study-sanctioned PC screening can be determined for all subjects. Under reasonable assumptions, and without elaborate modeling, applications of the method of Cuzick, et al (1997), will be used to provide unbiased estimates of the actual effect of PC screening.
Aims
#1a: Compare overall rates of gynecologic surgery and resultant morbidities between usual care and screened arms. We hypothesize that women in the screened arm had higher rates of surgery than women randomized to the usual care arm. #1b: Compare the rates of morbidity/complications associated with gynecologic surgery between the screened and usual care arms. We hypothesize that the rate of morbidity/complications is similar in both arms #2: Compare the rate of gynecologic surgery specific to the adnexa between the usual care and screened arms. We hypothesize that women in the screened arm had higher rates of surgery specific to the adnexa than women randomized to the usual care arm. #2b: Compare the rates of morbidity/complications associated with gynecologic surgery specific to the adnexa between the screened and usual care arms. We hypothesize that the rate of morbidity/complications is similar in both arms
