The ICU Care Knowledge Study

Family members have poor overall understanding of ICU patients' prognosis, resuscitation status and care, defined a priori as less than 75% correct responses to the yes/no ICU care questions.

Family members of all adult patients admitted to an intensive care of San Francisco General Hospital will be screened for enrollment and recruited sequentially over a period of nine months (August 1, 2005-March 31, 2006). We will determine the frequency of family members' correct responses to the survey questions about ICU care. Family member responses will be compared to the chart and real-time patient observation reality (gold standard).

Hypothesis # 2 Poor English language comprehension and low education level (defined as not having completed high school) are risk factors for poor ICU care understanding and knowledge.

During their surveys family members will be asked questions regarding their English language comprehension and education level. Association of these possible risk factors with low level of ICU care knowledge will be tested primarily using analysis of variance. Should our hypothesis #2 prove to be correct, our future goal will be to develop recommendations for interventions to increase understanding specific for these high risk groups. One of these interventions is described in Hypothesis #3 below.

Hypothesis # 3 Provision of the results of family members' knowledge assessment to health care providers will result in improved ICU care knowledge on subsequent testing.

Subjects will be randomized to two groups-one group's health care providers will be shown the results of family members' initial ICU knowledge assessment; the other group's providers will not be given these results. One to two days after providing this summary to providers, all randomized subjects will undergo a second interview assessing their ICU care knowledge. Changes in the frequency of correct responses between the two groups will be compared using mean differences in proportions with 95% confidence intervals.