I read with interest the article by Tsevat and colleagues [1] on health values in seriously ill patients. I agree that an assessment of individual preferences and values in health care is long overdue, especially as our medical interventions become more costly and scarce. Expensive therapies such as dialysis and mechanical ventilation should be reserved for patients who are comfortable with their intensity.

I am concerned, however, about the methods the authors used to study this issue. They used a time-tradeoff utility as a measure of a person's preference for quality (rather than quantity) of life. Patients stated how much of their remaining life expectancy they would trade for excellent health, and this Figure was assessed as the "value" of their lives. Although the authors did correlate this time-tradeoff with true values such as preferences for cardiopulmonary resuscitation and for life-extending care (even at the expense of pain and discomfort), this tradeoff still lacks much of the depth and substance of real life. A qualitative study may have provided richer data.

Many questions remain regarding the study's descriptive quantitative design. When personal values are being assessed, attention to the individual requires greater emphasis. For example, how concerned were the patients about the financial security of their surviving families? How did their religious beliefs affect their fight against or preparation for death?

These questions will not be answered by directing patients to pick a number such as a time-tradeoff utility. Also, some degree of bias is unavoidable when an investigator starts with his or her theory and then attempts to prove it. Physicians must therefore learn to conduct studies with more listening and fewer leading questions. Qualitative research designs offer this opportunity. In-depth, open-ended interviews with fewer patients are required. Theories could then be generated from the analysis of these interviews as trends are noted in the respondent's statements. These theories can then be validated using other populations.

Qualitative research designs are only appropriate for certain questions. For example, studies of most medications, interventions to decrease hospital costs, or an analysis of the efficacy of a new surgical procedure should be done using quantitative methods. However, attempts to determine an individual person's health care values or study the dynamics of the physician-patient relationship require a qualitative design.

The article by Tsevat and colleagues [1] does serve a useful purpose by showing that measures of patients' health care values widely vary. Large quantitative studies will not help us find information that is applicable to individual patients. Well-designed and -executed qualitative studies can answer questions that large quantitative studies cannot.