W-05.1: Do the Variables that Predict the Number of Cancer Clinical Trials in the United States Differ by Sponsor type?
Head Quality Analytics & Risk Management
Boehringer Ingelheim Germany
Using multivariate analysis, this study determined whether demographic characteristics of patients and/or the disease incidence rate are associated with the number of clinical trials available for participation. Data were examined by Sponsor type for 35 different tumor types.
Data collection included recruitment requirements (RR) from 1,156 oncology clinical trials from Clinicaltrials.gov and incidence rate and demographic characteristics of patients by tumor type from the 2013 Surveillance Epidemiology and End Results database.
Clinical trial recruitment requirements (RR) is a measure of the number of clinical trials available for cancer patient participation. These data are available by Sponsor type on Clinicaltrials.gov. INDUSTRY represents the pharmaceutical and biotechnology industry, GOV represents the U.S. Government, and OTHER represents entities such as non-profit organizations. The covariates AGE, RACE, and INCIDENCE were used in the models to determine group differences in the variables that predict the number of clinical trials offered for cancer patient participation. AGE is median age at diagnosis, RACE is the racial group with the highest age adjusted cancer incidence rate, and INCIDENCE is the adjusted estimated number of new cancer cases. The ANCOVA results showed that the association between the 3 covariates and RR significantly differs by sponsor type (F 5, 82) = 7.71, p < 0.0001). Linear regression models were run to determine variables that should be included in the best-fit models by Sponsor type. For INDUSTRY, INCIDENCE had a positive statistically significant association with RR; no other covariate reached statistical significance in the model (R2 = 0.2874). The R2 when only INCIDENCE was in the model was 0.2719. For GOV as with INDUSTRY, INCIDENCE had a positive statistically significant association with RR, no other covariate reached statistical significance in the model (R2 = 0.3142). The R2 when only INCIDENCE is in the model was 0.2988. For OTHER, both AGE and INCIDENCE were statistically significantly associated with RR; both associations were positive and RACE did not reach statistical significance (R2 = 0.3841). The R2 when INCIDENCE and AGE only were in the model was 0.3769. Across all sponsor types there is a statistically significant positive association between INCIDENCE and RR and for OTHER sponsor types, there is a statistically significant positive association between AGE and RR.
The literature shows an association between the number of clinical trials and both treatment advances and increase in 5 year survival rate for cancer patients. For this reason, it is important to understand which variables are associated with the availability of clinical trials for newly diagnosed cancer patients. Sponsors influence the number of clinical trials because they are the funders of clinical research. In this study, the outcome variable was RR which is an unadjusted measurement of clinical trial participation opportunities available to newly diagnosed cancer patients. It was hypothesized that because the funding mechanism and goal of the research differs depending on who is sponsoring the clinical trial, the factors associated with recruitment requirements might also differ by sponsor type. The ANCOVA analyses confirmed that there were group differences in the RR variable. A more detailed look by sponsor type followed with the regression analyses. The percent of variation in the outcome variable explained by INCIDENCE for all 3 sponsor types provides evidence that the number of newly diagnosed cancer patients is statistically significantly associated with the amount of clinical trial opportunities. For OTHER sponsor types AGE was also statistically significant in the model which means that for this sponsor type both AGE and INCIDENCE are significantly associated with cancer clinical trial availability. While the regression models indicated that 27%- 38% of the variation in RR could be explained by INCIDENCE (and for other sponsor type by INCIDENCE and AGE) further research in this area should be conducted to determine other predictors of the number of cancer clinical trials. Also, clinical trial availability is a necessary but insufficient condition for cancer patients actually participating in clinical trials, it is therefore important to look at factors that may be hindering clinical trial participation for newly diagnosed cancer patients.