Early methodological developments for clinical trials at the national heart, lung and blood institute


  • Max Halperin,

    1. George Washington University, Biostatistics Center, 6110 Executive Blvd. Rockville, MD 20852, U.S.A.
    Search for more papers by this author
    • Max Halperin wrote the first draft of this paper prior to his death in January, 1988. This paper was presented at the 1988 American Statistical Association annual meeting in New Orleans by the co-authors.

  • David L. Demets,

    1. University of Wisconsin-Madison, Department of Statistics, 1210 W Dayton, Madison, WI 53706, U.S.A.
    Search for more papers by this author
  • James H. Ware

    1. Harvard School of Public Health, 677 Huntington Ave., Boston, MA 02115, U.S.A.
    Search for more papers by this author


The National Heart Institute, now known as the National Heart, Lung and Blood Institute (NHLBI), initiated its first multicentre randomized clinical trials of rheumatic fever and rheumatic heart disease in 1951. The modern era of multicentre trials began, however, when the Coronary Drug Project was initiated in the 1960s. This trial and subsequent NHLBI trials stimulated a wide variety of research on clinical trial methodology. This paper reviews early methodologic developments in four areas. First, an organizational structure for multicentre clinical trials was developed and codified in the ‘Greenberg Report’ in 1967. Second, design considerations related to patient risk, non-compliance, a lag in treatment effect, and changing risk were explored. The ‘intention-to-treat’; principle was implicit in these investigations. Thirdly, the concept of periodic review of accumulating data, recommended in the Greenberg report, stimulated research on methods for sequential analysis. Three statistical approaches were developed and investigation of their statistical properties continues today. These approaches are usually described as group sequential, stochastic curtailment, and Bayesian methods. Finally, comparison of treatments in longitudinal studies has been an increasing part of NHLBI research and methods have been developed for design and analysis of longitudinal studies.