Although the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment—SUPPORT—came to public attention in the 1990s, it was conceived in the early 1980s, at a time when costs for high-technology medical interventions were increasing rapidly and people had begun to question the appropriateness of using all available measures to extend briefly the lives of people with untreatable, soon fatal, conditions. The Robert Wood Johnson Foundation, sponsor of SUPPORT and HELP, held a meeting in 1985 to discuss this. Subsequently, it asked Drs. William Knaus and Joanne Lynn to propose a study to improve the care of critically ill, hospitalized adults, specifically through improving the match between what patients wanted and the care they actually received.

A two-stage process was planned: Phase I was observational, and Phase II was a randomized trial testing an intervention tailored to address problems identified in Phase I. Planning, pilot testing, and recruitment took several years. Defining which patients would be eligible for the study was pivotal. The investigators chose conditions that were common and often fatal; that required important decisions during hospitalization; and that had stable treatment possibilities, to ensure that prognostic estimates would be similar throughout the study (this is one reason HIV/AIDS was not selected). Patients’ conditions had to be severe enough that about half would die within six months. The conditions selected were

  • acute respiratory failure,

  • chronic obstructive pulmonary disease,

  • congestive heart failure,

  • coma,

  • cirrhosis,

  • advanced colon or non-small cell lung cancer, and

  • multiorgan system failure with sepsis or malignancy.

Between 1989 and 1991, a full complement of 4,301 patients had been recruited to Phase I at the five large hospitals around the country that had been selected (out of 55 applications) as study sites. The following are key Phase I findings:

  • Patients with advanced life-threatening illnesses could be interviewed successfully about their treatment preferences.

  • Physicians often misunderstood patient preferences, especially when patients did not want high-technology, life-extending care.

  • Do-not-resuscitate (DNR) orders were often written very late—just before

of the studies, beyond this single finding, are hinted at by the 100 or so journal articles that have probed SUPPORT data (Phillips et al., 2000).

The failure of the planned interventions spurred the interested community to try to understand what went wrong and what could be done differ-

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement