Drug Information Association Logo

Time-Cost Optimization of Complex Clinical Trials

Time-Cost Optimization of Complex Clinical Trials

Background: A number of studies have been published presenting simulation models of the patient recruitment process. Patient accrual may be viewed in the context of the key components of the trial: study start-up, enrollment, patient stratification, globalization, and so on. Whereas previous research has tended to focus on the mechanics of the simulation models, we have focused on the key components of a clinical study and how to model each of the components. Methods: We modeled study start-up and patient enrollment processes using cumulative normal distribution functions for study start-up and mean cumulative functions to model cumulative enrollment per active site. The cost function was specified in terms of project time, number of sites and number of patients. We showed how the resulting time-cost model may be influenced by central institutional review boards, patient stratification, and the global distribution of sites. The model was fit to actual study performance data incorporating these factors. Sensitivity analyses were used to explore trade-offs between project time and cost. Decision parameters included the number of sites, which regions of the world to include, the mix of sites by region of the world, the relative size of various patient strata, mean cumulative enrollment rates, and so on. Mean and variance of the time to complete study start-up was sharply reduced for the set of sites using central institutional review boards in an outpatient cardiovascular trial. Patient stratification tends to prolong the trial and reduce the overall mean cumulative enrollment rates per active site. Mean enrollment rates tend to decline as strata fill up that are simpler to recruit into, and the total project time is prolonged. Including sites in Eastern Europe, South America, and India may increase mean relative enrollment rates by as much as three times the US rate and reduce total project time accordingly. Adding sites may reduce project time while adding only marginally to costs, because reductions in time-based costs offset additions in site-based costs. Monte Carlo simulation of the model enables probability statements regarding study completion time. Conclusion: time-cost models that reflect the complexity and idiosyncrasies of specific trials are critical to prudent planning for individual trials and clinical development plans. Given the formidable challenges in attracting funding for novel drugs and devices, time-cost models have an important role to play in planning and managing global clinical trials.