Various models of community research have been employed to assess community health needs and address community health problems. Infant mortality remains a problem in urban minority communities throughout the United States. This article compares various community research models with a three-tiered community involvement model used in Washington, D.C., an urban, predominantly African American community with one of the worst rates of infant mortality in the nation. The research design incorporated community input throughout the research design process, resulting in a better understanding of issues and problems associated with infant mortality within the community setting. The use of the model improved community participation in the research process, facilitating improved health initiatives.

Computer interviewing to obtain sensitive information is not a new concept. However, concerns about the acceptance of computers in disadvantaged populations with potentially low literacy led us to combine audio- and touch-screen technologies with an audio computerized self-report interview to obtain information about alcohol use. This study evaluated acceptance and ease of use by a disadvantaged population of pregnant women in the District of Columbia. Patients attending an initial visit at prenatal clinics answered questions anonymously about their consumption of alcoholic beverages and other personal information. The questionnaire was programmed on a laptop computer. The computer administered the recorded questions via earphones, as well as displayed them on the screen, and patients answered by touching the computer screen. Results were immediately available. A total of 507 women were interviewed, who were primarily African American, non-Hispanic, and never married. Nearly 24% did not complete a high school education, 43% were unemployed, and 30% received public assistance. Most of the women (59%) used computers occasionally (a few days a month) or never. Nearly all patients (96%) reported that the computer was not difficult to use, and approximately 90% liked answering the questions by computer. The study demonstrates that using computers to screen for alcohol use in disadvantaged pregnant populations is feasible and acceptable to the patients.

Purpose: This study investigated the use of the TWEAK and nine alternative screeners for predicting high-risk and moderate-risk drinking during pregnancy. Method: The analysis was based on self-reports from 404 lifetime drinkers who presented for an initial visit at nine prenatal clinics in Washington, DC. Data were collected anonymously by having women directly enter their responses onto an audio, computer-assisted interview that was programmed onto a laptop computer. Pregnancy risk drinking status was based on both average daily volume of intake and frequency of drinking 3+ drinks in a day. Each of the alternative screeners was constructed by adding one additional risk indicator to the TWEAK, and three different scoring options were explored. Results: Using thresholds of 2 points for high-risk drinking and 1 point for moderate-risk drinking, the TWEAK demonstrated a sensitivity and specificity of 70.6% and 73.2% for high-risk drinking and a sensitivity and specificity of 65.6% and 63.7% for any (high- or moderate-) risk drinking during pregnancy. None of the alternative screeners resulted in significant improvement, but the addition of current smoking status showed enough promise to warrant further testing in larger samples. Conclusions: Despite some loss in sensitivity and specificity, the TWEAK, in its original or a modified form, can be extended to measures of high-risk drinking that incorporate infrequent heavy intake and can be used to test for moderate- as well as high-risk drinking. Because identification of moderate-risk drinkers substantially increases the pool of women targeted for intervention, cost implications must be considered in designing appropriate interventions.