Cost-effectiveness analyses (CEAs) are disproportionately conducted on pharmaceuticals relative to medical and surgical procedures. In order to better understand the possible reasons for that discrepancy, IVI embarked on a project to compare the key characteristics of pharmaceutical versus medical/surgical CEAs and the underlying effectiveness studies on which they rely. We wondered about the importance of industry versus government sponsorship, the availability and use of randomized controlled trial (RCT) versus observational data, the importance of property rights, the use of advanced statistical techniques for dealing with observational data, and the degree to which CEAs conformed with the recommendations of the Second Panel on Cost-Effectiveness in Health and Medicine.

To answer these questions, IVI collected data to augment existing data on CEAs from the Tufts Medical Center Cost-Effectiveness Analysis Registry. Specifically, the project collected and examined information on the following:

  • Source of data used to measure effectiveness
  • Sponsorship of the CEA and underlying effectiveness studies
  • Statistical techniques used to measure effectiveness
  • Property rights for the intervention being studied
  • Other issues of interest, including time horizon, whether costs outside of the health care sector and other CEA elements are considered in the analysis, and degree of adherence to current CEA guidelines

We found that pharmaceutical CEAs and effectiveness studies were primarily sponsored by industry and were far less often sponsored by government. Studies that mixed both pharmaceutical and medical/surgical interventions saw the most government sponsorship. Pharmaceutical studies predominantly used RCT data, while medical/surgical studies predominantly used observational data. Advanced statistical techniques for dealing with endogeneity or unmeasured confounder bias were seldom used with observational data. The existence of property rights and industry sponsorship increased the likelihood of using RCT data, even on the subset of medical/surgical studies.

The findings suggest that government sponsorship is important for maintaining or increasing the number of effectiveness studies on medical/surgical procedures. Efforts to increase RCTs on medical/surgical procedures or the use of statistical techniques to deal with endogeneity bias in observational data may be in order.


To the greatest extent possible, IVI makes all research materials publicly available.

The following materials are currently available for download here:

  • Research Protocol: Reviewer Comments and Responses
  • Research Protocol: Final Protocol
  • Complete Dataset and Data Dictionary (IVI-extracted data only; does not include variables from Tufts CEA Registry)
  • Draft Report: Reviewer Comments and Responses
  • Final Report