Success rates of American clinical oncology trials by geographic factors

We aim to evaluate relationships between the locations and success rates of United States oncology clinical trials by analyzing geographic features such as median income and proportion of practicing oncologists by ZIP code. A dataset of 15,658 trials with at least one site in the 50 states or the District of Columbia was collected by integrating trial outcomes curated in Trialtrove, locations from ClinicalTrials.gov, incomes from the U.S. Census, oncologist locations from the Center for Medicare Services and other sources. We found that ZIP codes in which trials are conducted are skewed away from lower median incomes (p = 4.05e-08, Kolmogorov-Smirnov test). When ZIP codes were partitioned into lower, middle and upper terciles of median income, success rates by ZIP code were significantly lower in the low-income range relative to the middle- and higher-income ranges (lowest vs. highest p < 0.0001, OR 1.13 [1.10-1.16]). Trials conducted in more ZIP codes are significantly more likely to succeed than trials conducted in fewer ZIP codes. We identified candidate ZIP codes with sufficient patient populations and practicing oncologists to support clinical oncology trials but have never had a trial previously.