Login to Access Video or Poster Abstract: MP47-05
Sources of Funding: This work was supported by David H. Koch provided through the Prostate Cancer Foundation; the Sidney Kimmel Center for Prostate and Urologic Cancers; SPORE grant from the National Cancer Institute to Dr. H. Scher (grant number P50-CA92629); and a National Institutes of Health/National Cancer Institute Cancer Center Support Grant to MSKCC (grant number P30-CA008748). None of the funding sources had involvement in the conduct of the research or preparation of the manuscript.

Introduction

In many areas of prostate cancer research, such as comparing patient outcomes for radical prostatectomy versus radiotherapy for localized cancer, there are a limited number of randomized trials. Observational studies are used to address such comparisons. Instrumental variables analysis is a methodology to control for confounding in observational studies. Geographic area is being used increasingly as an instrument. We conducted a literature review to determine the properties of geographic area in studies of cancer treatments.

Methods

We reviewed the literature to identify cancer studies performed in the United States which incorporated instrumental variable analysis with area-wide treatment rate within a geographic region as the instrument. We assessed of the degree of treatment variability between geographic regions, assessed control of confounding afforded by geographic area and compared the results of instrumental variable analysis to those of multivariable methods.

Results

Geographic region as an instrument was relatively common, with 22 eligible studies identified, many of which were published in high-impact journals. Prostate cancer studies made up nearly half of the eligible studies found (10 of 22), including studies of surgery vs. conservative management, primary androgen deprivation vs. conservative management, surgery vs. radiotherapy and open vs. robotic surgery. Geographic region was only weakly associated with the intervention. Most studies reported an absolute difference in treatment rates between high and low use areas of 5% to 20%, with the largest difference being 31%. Absolute differences between high and low use areas reported in prostate studies ranged from 8% to 22%. Four out of seven of these studies reporting on covariate balance found at least one covariate to be associated with treatment prevalence. Eight out of eleven studies including five out of six prostate cancer studies found statistically significant effects of treatment on multivariable analysis but not for instrumental variables, with the central estimates of the instrumental variables analysis generally being closer to the null.

Conclusions

We recommend caution when using geographic region as an instrument in observational studies of treatments for prostate and other cancers. The value of geographic region as an instrument should be critically evaluated in other areas of medicine.

Funding

This work was supported by David H. Koch provided through the Prostate Cancer Foundation; the Sidney Kimmel Center for Prostate and Urologic Cancers; SPORE grant from the National Cancer Institute to Dr. H. Scher (grant number P50-CA92629); and a National Institutes of Health/National Cancer Institute Cancer Center Support Grant to MSKCC (grant number P30-CA008748). None of the funding sources had involvement in the conduct of the research or preparation of the manuscript.