Individualized PHRs were developed for patients receiving care at the adult HIV/AIDS Clinic at SFGH 
. The PHR populates automatically in real time with selected data from the clinician's entries in the patient's Electronic Medical Record (EMR). Any HIV/AIDS-infected patient who received primary care at this Clinic between July 2009 and June 2010 and provided written informed consent was eligible for the parent RCT. The goal of this trial was to assess the effect on HIV-related health outcomes of web-based access to personal (PHR) or general (non-PHR) health information via individualized web-based accounts.
Recruitment, Consent and Enrollment
During clinic sessions, Research Assistants approached patients individually in the clinic's waiting room, provided them with a general overview of the study's goals and requirements, and invited them to participate. Patients were advised that they would be asked to complete web-based study surveys at 0, 6, and 12 months' follow-up and would be compensated five dollars in gift certificates to a major grocery chain. A Research Assistant met individually with each participant in a private office to complete the enrollment process and to obtain consent.
Minimal Instruction in Computer Use and Internet Access
Given the safety-net setting of the RCT, Research Assistants helped participants obtain a health information account, explained general and study-specific aspects of a web-based health information account, and offered rudimentary training in computer use – such as how to use a mouse, how to “click and drag,” how to open an Internet browser, and how to complete the surveys embedded within the account. For subsequent access, participants were referred to public libraries and other locations that offer free Internet and computer use where they could complete their study-related surveys.
Enhanced Instruction in Computer Use and Internet Access
A need to measure computer skills was not anticipated prior to the RCT. However, during accrual months 1–3, Research Assistants noted wide variation in participants' computer abilities and began to record each participant's level of computer competency (beginner, intermediate, or experienced). Study procedures were modified during accrual months 4–12 to have Research Assistants coach participants to complete baseline surveys at the clinic during their initial web-portal access, without guiding responses or breaching confidentiality, and to record the time they spent helping the participant.
Data Collection and Analysis
Our two dichotomous outcomes, RCT participation by clinic patients and survey completion by RCT participants, were extracted from clinic records and study participation computer logs. They corresponded with distinct research questions and were analyzed separately. In both cases, the primary variable of interest was presence/absence of a MH and/or SU condition. MH and SU conditions were diagnosed and recorded by primary care providers in the EMR; nicotine use was not studied. When these conditions were not recorded, we assumed they were absent.
Among clinic patients, we addressed “willingness to engage” via consent (yes/no) to participate in a randomized trial that provided access to health information via individualized web-based accounts. We used a multivariable logistic regression model to estimate the association between participation in the RCT and MH/SU status, adjusted for demographic characteristics. We analyzed age in decades, combining individuals 60 and older.
Among randomized participants, we addressed “ability to engage” via completion (yes/no) of all nine baseline surveys embedded in the web-based accounts. We used a multivariable logistic regression model to estimate variation in survey completion by MH/SU status, on-study training in computer use (two levels), computer competency (four levels), and demographic characteristics. To identify differential training benefits, we included in the model interactions of training strategy with MH/SU status and with level of computer competency. We summarize these effects via odds ratios (OR), 95% confidence intervals (CI), and likelihood ratio (LR) p-values.
This study was approved by the University of California San Francisco Committee for Human Research (#0833272) and the Privacy Board of the San Francisco Department of Public Health. The randomized controlled trial was registered at clinicaltrials.gov, Registry No. NCT00972348.