Understanding Site Performance Differences in Multinational Phase III Clinical Trials

Abstract

Background

This paper explores the impact of various factors on completion times in phase III clinical trials; specifically, why some sites are able to complete more patients during the course of the clinical trial than other sites, including those within the same trial.

Method

Seven companies provided research data for inclusion in this analysis. The study sample inclusion criteria were the indications of asthma, chronic obstructive pulmonary disease, depression, schizophrenia, bipolar disorder, hypertension, type 2 diabetes mellitus or menopausal syndrome; phase III; study completion between 2002 and 2005; a New Chemical Entity for an initial (primary) indication and formulation; adult outpatient subjects; and multinational design, with sites in the US, at least one Western European country, and at least one Eastern or Central European country. The analysis assessed both study-level and site-level variables.

Results

This paper analyses the results from 262 phase III studies, involving 2047 sites, meeting the stated inclusion criteria. Unsurprisingly, the number of enrolled patients is a clear predictor of the number of completed patients (standardised beta coefficient 0.877; p=0.001). Better performing sites started enrolling patients earlier than other sites and continued to enrol patients over a longer period of time. The time between when a site enrols its first patient and the time the first site in that study enrols its first patient, is a specific predictor of how well a given site will perform; indeed the relationship is inverse so that the greater the time difference, the poorer the performance. Sites involved in studies using competitive enrolment and milestone payments generally complete a higher number of patients. The same holds for sites in clinical trials using upfront, nonrefundable start-up payments, although the relationship is weaker. Interestingly, higher grant payment levels do not translate into better site performance. Site level variables were consistent with study level variables. In addition, investigators with more recent clinical research experience with that sponsor company perform better and sites using a central institutional review board (IRB) outperform sites using local IRBs or a combination of central and local IRBs.

Conclusions

Three major conclusions emerge from the analysis. First, better performing sites begin enrolling quickly, providing more time to enrol patients, and these sites continue to enrol and complete patients over that extended period of time. Second, investigators with overall clinical experience, especially with the sponsor company, are usually the better performers. Third, how much an investigator is paid seems less important than how that investigator is paid. Higher payment levels are not associated with improved site performance. In contrast, the use of competitive enrolment and milestone payments does seem to positively affect performance.