Two randomized controlled trials were conducted that examined different methods of rapid HIV testing information delivery. The trials were conducted sequentially. For both trials, participant understanding of rapid HIV testing fundamentals was evaluated using a questionnaire. The first trial compared participants who received no information to those who underwent an in-person discussion. The second trial compared participants who underwent an in-person discussion to those who watched an educational video. The Rhode Island Hospital institutional review board (IRB) approved the study. This study was a pilot to a larger trial comparing the video to the in-person discussion information delivery methods among patients undergoing rapid HIV testing with OraQuick^®.

The two trials were performed at the Rhode Island Hospital ED, which is a large, urban, academic ED in the northeastern US. Patients eligible for the trials were ED patients 18-55-years-old who were not pregnant; not incarcerated or in home confinement; not known to be HIV infected; able to speak, read, and write in English; were not deaf; and were not critically ill, intoxicated, or presenting to the ED for a psychiatric problem. Participants could only be members of one trial.

Trial One was conducted February-April 2005 and Trial Two between April 2005-January 2006. A research assistant (RA) conducted the trials weekdays during arbitrarily selected eight-hour shifts. The shifts began between 7 a.m. and 3 p.m. and ended between 3 p.m. and 11 p.m. For each trial, a research assistant (RA) randomly screened for study eligibility 50% of the medical records of patients who were present in the ED during the RA's shift. Before each shift, the RA used a random number generator to choose either the number 1 or 2 11. These numbers represented "odd" and "even" terminal medical record digits, respectively. The RAs screened only the "odd" or the "even" medical records for that shift as determined by this random selection scheme. The random selection was conducted to reduce the possibility that the RA might selectively choose patients to include in the study that could affect the outcome of the study, i.e., reduce selection bias.

The RAs screened the ED medical records in the ambulatory care and urgent care areas of the ED looking for patients who potentially met the study eligibility criteria. Patients in the critical care, substance abuse, and psychiatric evaluation areas of the ED were not screened for study inclusion. The demographic profile (age, gender, race/ethnicity, marital status, and insurance type) for every patient screened was recorded into the study database. The RAs also recorded the reasons for exclusion from the study and did not approach those who were ineligible by ED medical record review.

Patients who appeared to meet eligibility criteria by medical record review were approached by the RA. The RA briefly outlined the study purpose to the patient and asked for their verbal consent to confirm their demographic profile, study eligibility, and to ask additional questions about their HIV testing history and HIV status. Patients who declined to answer these questions or who were found to meet exclusion criteria during this brief interview (e.g., told the RA that they were HIV-infected) were excluded from the study. The RA recorded the reasons for their exclusion. Patients who met study eligibility were apprised of the study and invited to participate. Verbal agreement to participate was obtained.

In Trial One, patients were randomized to an in-person discussion on rapid HIV testing information arm or a no-information arm. The no-information arm received no rapid HIV pre-test information of any type. In Trial Two, patients were randomized to an in-person discussion arm or a video arm. The RA randomized patients for each trial using their penultimate medical record number. Odd-numbered patients were assigned to the in-person discussion while even-numbered patients were assigned the other arm, respective to the trial. Following the delivery of rapid HIV test information (or no information), the RA orally administered an identical questionnaire to all participants to assess their comprehension on the material presented. At the completion of the questionnaire, respondents were given an opportunity to discuss any questions or concerns about the material or topics presented or the study itself. At the completion of the study, participants were provided a copy of the written consent summary and a brochure on HIV and HIV testing that also gave a list of locations in the local area where they could be tested for HIV.

For both trials, the in-person discussion was a semi-scripted conversation between the participant and RA on five topics: HIV and AIDS, HIV transmission, HIV prevention, HIV testing, and rapid HIV testing. An outline of the points covered in the in-person discussion is provided in Figure 1. The outline was created by the study authors and is based upon CDC recommendations for pre-test information 8 as well as manufacturer details on rapid HIV testing with OraQuick^® 12. RAs conducting the study memorized the outline and were trained to deliver it through mock-interviews. The principal investigator observed the RAs during the trials, critiqued their performance, and corrected deviations from the outline. The RAs also underwent training by the state department of health to be certified as HIV test counselors.

The in-person discussion resembled a didactic presentation yet was conducted in a conversational style. Participants were asked an opening question for each of the five topics. The RA would endorse or correct the participant's response then state the points relevant to that topic---whether or not the participant answered the opening question correctly. To ensure uniformity of the presented information, the RA limited the discussion to the outline. The RAs had an abbreviated version of the outline in Figure 1 to use as a reference during the in-person discussion, but were not permitted read the outline verbatim to patients.

In Trial Two, the video information arm patients watched the 9.5 minute animated and live-action educational video "Do you know about rapid HIV testing?"10 that contained the same rapid HIV pre-test information elements presented to the in-person discussion arm. The video and its development are described in the accompanying manuscript: Development of the rapid HIV testing video, "Do you know about rapid HIV testing?" [see Additional file 1]. Participants randomized to the video arm watched the video on a hand-held Tablet PC while awaiting medical care in the ED. They listened to the video using headphones provided by the RA.

Participants in both trials were asked to complete the "HIV pre-test information comprehension" questionnaire to assess their comprehension of rapid HIV testing fundamentals. The development of the 26-item, "true", "false", "I don't know" questionnaire is described in the accompanying manuscript: Development of the "HIV pre-test information comprehension" questionnaire [see Additional file 2]. Participants in both trials received the identical version of the questionnaire. The RAs administered the questionnaire verbally to participants and recorded the responses onto a Tablet PC database. RAs provided all participants with a standardized, brief introduction to the nature of the questions, the reason for the questionnaire ("to determine how well we were able to give you this information about HIV and rapid HIV testing"), and the form of the responses needed ("true", "false", or "I don't know"). The RAs also informed participants that they would not be able to assist in answering the questions during the administration of the questionnaire, but could repeat questions as needed. Participants were asked to choose the response they thought was best, but to feel free to answer any question as "I don't know" when unsure of an answer. The RAs discussed the questions and answers with the participants after they completed the questionnaire.

The RAs recorded all study data (demographic and HIV test history, eligibility determinations, responses to the questionnaire, etc.) onto a Tablet PC study database using the QDS^® (NOVA Research, Bethesda, MD) data entry and management software. To ensure accuracy of the data entry, the data were entered in duplicate with immediate data verification. The data were analyzed using Stata 9.2 (Stata Corporation, College Station, TX). The analysis included tabulating the results of the medical record screening and eligibility determinations; participant demographic profiles, educational level, and HIV testing histories; and responses to the "HIV pre-test information comprehension" questionnaire. Patients were compared by their demographic profiles according to their eligibility, participation, and study arm assignment using Pearson's χ² test for categorical values; Fisher's exact test for categorical values with small samples; two-sample tests of binomial proportions for dichotomous values; and Wilcoxon rank-sum test for continuous values. Participant educational level and HIV testing history were compared by study arm assignment using Pearson's χ² test and Fisher's exact test.

The mean score and standard deviation, median, and range for each of the five topics areas of the questionnaire and the entire questionnaire were calculated by arm assignment for both trials. We originally estimated requiring a sample size of 55 for each arm of Trial Two to demonstrate a 12% absolute increase in mean scores. We assumed the same standard deviation for each arm, a normal distribution of scores, α = 0.05, β = 0.10, and use of Student's t-test. However, using plots of the scores for each trial and the Wilk-Shapiro test, we observed that the scores were not normally distributed, therefore, the scores for the arms were instead compared using the Wilcoxon rank-sum test. Differences were considered significant at the α = 0.05 level. In an exploratory analysis, the proportions of participants answering each question correctly were compared by arm assignment using two-sample tests of binomial proportions. This exploratory analysis was used to help identify sub-topics that might not have been well addressed in the pre-test information stage or highlight questions that might not have been well understood by participants.

Table 3 shows by arm assignment the percentages of participants correctly answering the "HIV pre-test information comprehension" questions. The results of the two-sample tests of binomial proportions are provided in an exploratory analysis to identify potential sub-topics or questions that might not have been well understood by participants. Across both trials, there were twelve questions correctly answered by more than 50% of the patients in all arms and five correctly answered by more than 75% of patients in all arms. Fewer than 50% of participants in all arms were able to correctly answer the questions regarding the role of HIV antibodies (question 15) and phlebotomy for the rapid HIV test (question 25).