Factors Associated with Multiple FDA Review Cycles and Approval Phase Times

We examined the regulatory review histories of 298 approved new drugs and biologics with new drug applications (NDAs) or new biologic license applications (BLAs) submitted to the US Food and Drug Administration (FDA) during fiscal years 1996–2006 for factors that were associated with multiple review cycles and with longer or shorter total approval phase times (sum of FDA action and sponsor response times) for a given number of review cycles. Using logistic regression models, we found that the likelihood of multiple cycle review varied by therapeutic class, product type (odds ratio = 3.1 for BLAs), fast track status (odds ratio = 0.3), limited sponsor regulatory experience (odds ratio = 2.1 for first US approval), and more recent submissions (odds ratio = 1.7 for fiscal year 2000–2003 submissions compared to fiscal year 1996–1999 submissions). We also applied least squares multivariate regression analysis to explain the variation in total approval phase times with a number of drug, sponsor, and regulatory characteristics. After controlling for the number of review cycles and other factors, we found longer average approval phase times for more recent submissions (56% longer for priority rated compounds with multiple cycle reviews and submissions during a later period, P < .0001), compounds approved in CDER versus CBER (18% longer for single-cycle NDAs versus single-cycle CBER approvals, P = .0372; 35% longer for single-cycle BLAs approved by CDER versus single-cycle CBER approvals, P = .0264; and 44% > longer for multiple-cycle BLAs approved by CDER versus multiple-cycle CBER approvals, P = .0209), compounds with December submissions (18% longer for compounds with multiple cycle reviews, P = .0069), and compounds with advisory committee meetings (13% longer for compounds with a single review cycle, P = .0344). We found shorter average approval phase times for compounds with a fast track designation (21 % shorter for compounds with multiple cycle reviews, P = .0825) and accelerated approval status (24% lower for compounds with a single review cycle, P = .0010). Statistical models such as those presented here can help improve the precision with which drug developers and regulators predict approval times and assess the effectiveness of regulatory programs designed to speed the drug approval process.