Overlap Among VACS 3, VACS 5, and VACS Samples
This report summarizes VACS 3, VACS 5, and the “full study” or VACS. As can be seen on the timeline of enrollment and follow-up (Unique patients, ), there is overlap. As of this writing and taken together, the VACS 3, VACS 5, and full studies have enrolled and will follow for long-term outcomes more than 7200 unique veterans (3661 HIV+, 3608 HIV−) from 9 sites. (One site included in VACS 3 has not participated in subsequent studies, ie, Cleveland, Ohio.) The full study (VACS) includes 83% of all these subjects (81% of the HIV+, 84% of the HIV− veterans). Considering only HIV infected subjects: 196 of 881 (22%) of VACS 3 subjects and 861 of 1031 (84%) of VACS 5 subjects enrolled in the full study. Because the full study will continue to enroll, these numbers will change over time.
VACS 3 and VACS 5
The 3 site study (VACS 3) has been described in detail2
and is included, with the other studies, in . VACS 5 included 1 patient and 1 provider survey as well as full access to each participant’s electronic medical record. It was conducted at 5 VA medical centers (Atlanta, GA; Bronx, NY; Los Angeles, CA; Manhattan/Brooklyn, NY; and Houston, TX). Because it was designed as a feasibility study for the full VACS, the study’s methods used for establishing target enrollment and consenting subjects were the same as described herein.
From September 2001 through June 2002 veterans (1031 HIV+; 740 HIV−) were enrolled. HIV-infected participants in VACS 5 both had a median age of 49 years and were 99% male. Nonparticipants were more likely to be black, have fewer clinic visits, higher viral loads, and to have been on highly active antiretroviral therapy (HAART) for a shorter period of time (P < 0.05).
VACS 5 patients participated in 2 pilot substudies: neurocognitive and psychiatric testing (n = 50) and blood sampling and banking (n = 254). The neurocognitive and psychiatric testing portion of the study has been described in detail.3
The blood and DNA banking involved asking patients to donate samples of blood for future testing of other laboratory markers not commonly ordered during routine visits. Tests conducted on these samples included iron, vitamin B12, folate, homocysteine, mitochondrial AST in serum, and mitochondrial DNA damage. These results will be reported separately (manuscripts in preparation).
VACS (Full Study)
Before enrollment began, all study sites received approval for the use of human subjects by their respective Internal Review Boards (IRBs) as well as overall approval for the VACS Coordinating Center, initially located at the Pittsburgh VA. In July 2003, the Coordinating Center relocated to the VA in West Haven, Connecticut. Principal Investigators, study coordinators, data analysts, and all personnel associated with this project have undergone required training in human subjects protection, good clinical practices, privacy (or Health Insurance Portability and Accountability Act [HIPPA]) and conflict of interest. New study coordinators at each site are trained by an experienced VACS coordinator.
VACS opened to enrollment in June of 2002 initially at the same aforementioned VACS 5 sites. In February of 2003, 3 new sites were added: Baltimore, Maryland; Pittsburgh, Pennsylvania; and Washington DC. Patients are enrolled from the infectious disease (HIV-infected) or general medical clinics (age-, race-, gender-, site-matched uninfected comparators). In either clinic, patients are assigned to a single primary care provider who may have been a general internist, an infectious disease specialist, a physician assistant, or an advanced practice registered nurse. Baseline enrollment for reporting purposes is defined as lasting from June 2002 to September 2004. During this interval, 5998 subjects enrolled (2979 HIV+, 3019 HIV−). The study continues to enroll new HIV infected subjects and age/race/site-matched uninfected comparators to replace those who have died or are lost to follow-up to avoid becoming a “survival cohort” and requires an annual enrollment of approximately 5% of the overall sample. Survival cohorts occur when samples are identified in 1 period and then followed for an extended time without further enrollment. A survival cohort no longer represents the target population because individuals are required to have lived an extended period after presentation to be included in the sample.
Each site is monitored by a Principal Investigator, Co-Principal Investigator, and study coordinator. The study coordinator recruits HIV-infected participants from the infectious disease clinics and an age/race/site-matched HIV-uninfected comparators from the general medicine clinics. The study is explained to the participants in detail, and once written informed consent is obtained, the participants are asked to complete a questionnaire before leaving the clinic. The participants’ providers are asked to complete a shorter questionnaire that contains questions specific to that patient on the day of his/her appointment and a one-time provider characteristic questionnaire which asks for details on the provider’s practice, caseload and training. The questionnaires are developed using Teleform software (Verity, Sunnyvale, CA), which allows investigators to both design surveys and enter data collected via scanning technology.
Using the EMR, we determine the number of patients who are deceased or lost to follow-up on an annual basis. For purposes of reporting, we define lost to follow-up as those who have not been seen within the VA system for 12 months. Sometimes these patients return to the study in subsequent follow-up. Further, no veteran is truly “lost to follow-up” with respect to mortality endpoints. We eventually achieve mortality endpoints even on patients considered “lost” because of the availability of a death benefit paid to defray funeral expenses for all veterans.4,5
In a 12-month period, approximately 2.5% of the HIV-infected veterans (not known to be dead) who received VA care in the previous 12 months will not receive it in the subsequent 12 months. For HIV-negative comparators, the proportion is 4%.
The first annual survey follow-up occurred from September 2003 to September 2004. The VA sites in Baltimore, Maryland, and Washington, DC, did not participate in this initial follow-up as baseline enrollment was still underway. The second follow-up occurred from October 2004 to September 2005. A third wave of follow-up surveys is currently underway. All sites have also begun enrolling VACS patients in the blood and DNA banking substudy.
Target Sample Calculation
Using the VA Immunology Case Registry of all HIV-infected veterans in care,6
the number receiving HIV care was determined by VA site and clinic. Target enrollment was set at 65% of HIV-infected patients receiving care at each Infectious Disease clinic in fiscal year 2002 and was stratified based on age and race for each site. Targets for controls were set for each site to match the HIV infected site enrollment targets by age and race.
Once permission from the providers is obtained, study coordinators approach patients as they arrived at the clinics for their regular appointments. The coordinators describe the nature of the study and obtain informed consent. Patients are given the option, if eligible, to enroll in the telephone interview substudy, which requires a separate consent form. Patients are then given a paper questionnaire to be completed before the end of the clinic visit and are paid $20 for their participation. By signing the informed consent, participants give their permission for study investigators to review all electronic medical records. Patients also grant permission to be recontacted in the future.
With few exceptions, standardized instruments are used in each questionnaire ().7–35
There are items concerning medical conditions, health (such as height, weight, frequency of exercise, smoking), as well as questions regarding homelessness, income, health care utilization, satisfaction with care, health-related quality of life, medical care and insurance outside the VA, employment and education, and race/ethnicity, gender, and date of birth. Most of these items were also used in the Veterans Health Survey, a national sample of all eligible veterans.9a
We adapted these measures to allow benchmarking against this representative sample of veterans. The questions for health conditions are expanded in the infectious disease questionnaire to include AIDS-defining conditions using the Centers for Disease Control and Prevention list.11
VACS Survey and Other Supplementary Data Items
The surveys also include detailed information regarding substance use (alcohol, cigarettes, and both illicit and prescription drugs of abuse; ). These items attempt to capture details of current patterns of use as well as prior use and whether use was discontinued due to health problems. Additionally, information concerning readiness to change current substance use behaviors is also being collected for alcohol, cigarette, and drug use.
Other domains that are collected by self report include social aspects of health, social contacts and coping, religiosity and spirituality, medication adherence, symptom burden, symptoms of depression and anxiety, health-related quality of life, and functional status (). In addition, participants from the general medicine clinics are asked whether they have ever been tested for HIV and if they think they might be at risk. Participants from the infectious disease clinics are instead asked when they had their first positive HIV test and how many months had elapsed until they sought treatment.
Two separate provider questionnaires are administered (); the first is a one-time provider characteristic questionnaire, which is given to both general medicine and infectious disease providers. This questionnaire is designed to provide information on provider demographics, specialties, degrees, years in practice and caseload makeup and is described in detail elsewhere.20
The other provider questionnaire is administered to coincide with the patient’s visit and was developed to provide information from the provider about that particular patient. This questionnaire is given annually at all follow-up visits. General medicine and infectious disease providers received separate questionnaires; in the full study, the infectious disease version was expanded to include questions concerning HIV conditions and medications.
Death and Cause of Death
On a quarterly basis, we conduct searches of the Beneficiary Identification Records Locator Subsystem (BIRLS, a benefit paid to defray the costs of burial) and inpatient files in the EMR to determine whether any VACS subject has died. This approach has been favorably compared with the use of the National Death Index.5
Similarly, study coordinators also report deaths directly to us. Once a death has been identified, the study coordinator is notified and he/she collaborates with the primary care provider to complete a cause of death form, which provides information on date and cause of death, if known. These forms and supporting electronic medical record data are reviewed by a standing “Cause of Death Committee” to insure that the forms are completed in a consistent manner. The form used and the manner of review completely parallel that described by the Collaborations in HIV Outcomes Research-US cohort.7
Blood and DNA Sampling
VACS is working in partnership with the Massachusetts Veterans Epidemiology Research and Information Center in Boston, Massachusetts, and the Palo Alto Cooperative Studies Program Coordinating Center in Palo Alto, California, to coordinate a major blood and DNA banking substudy. All of the HIV-positive and 50% of HIV-negative patients already enrolled in VACS will be asked to donate a sample of blood for further testing (target is 80% of the enrolled HIV-infected veterans and 40% of enrolled HIV-negative veterans). The protocol for blood banking encompasses prospective assays not routinely performed at the VA. Initial tests will include those to determine evidence of liver injury, cardiac injury and bone marrow function. Additional assays on banked specimens will involve studies related to alcohol, aging, comorbidity and drug toxicity and will be based upon a formal review process.
In addition, VACS in conjunction with VA Cooperative Studies DNA Bank36
will store DNA and plasma from consenting patients. The DNA will be used for future studies of genetic factors associated with HIV, HIV treatment, alcohol use and comorbid conditions. Proposals for these studies will be reviewed under a separate process.
To capture in-depth information on alcohol use, abuse and dependence, and medication adherence, participants in the full study were asked to participate in a telephone interview. Participants were eligible if they had consumed 1 drink during their lifetime. Patients who refused the telephone interview were still permitted to continue their participation in the full study.
Telephone interviews were conducted by experienced interviewers at the University Center for Social and Urban Research at the University of Pittsburgh. The interviewers were qualified by senior research investigators in the standards of best practices specific to the alcohol instruments used in the interview. Additional interviewer training included role-playing, mock interviewing, and qualifying interviews to insure optimal skill and technique. The baseline telephone interview substudy has been described in detail elsewhere.8
A subset of the participants in the baseline telephone interview were called 1 year later for a second interview (n = 1173, 77% of target). This second interview used items from the Composite International Diagnostic Interview (CIDI) and items modified from the Substance Abuse Module (SAM).9
Data are currently in the analysis phase. Future plans include a validation of the TLFB in a subset of patients at 4 of the 8 study sites.
During the first year of baseline enrollment for the full study, 16 patient focus groups were conducted at all 8 VAMC study sites (total n = 147). The primary purpose of this substudy was to gather qualitative data that would inform the development of behavioral interventions to improve outcomes for patients, particularly related to alcohol use and management of their chronic diseases. These groups consisted of separate groups of approximately 8–12 General Medicine (HIV uninfected) and Infectious Disease (HIV infected) study participants.
Participants in the patient focus groups were paid $25 for their time. Discussion explored patient beliefs, attitudes, experiences, and behaviors relating to alcohol and substance use and abuse and their impact on chronic disease, including HIV infection. The particular focus was on positive behavior change; other health behaviors were explored, such as adherence to medications and healthy lifestyle choices as they affect outcomes. This qualitative data will be reported separately and will be used to inform the design of a clinical intervention to help patients reduce or abstain from alcohol use or abuse and to adopt healthy behaviors.
During the second year of the full study, provider focus groups were held at all study sites. A total of 47 infectious disease providers and 53 general medicine providers participated. Topics for discussion included: ways of helping patients change behavior, support services at the VA, and the use of computer reminders. Providers were not paid for their time.
All focus groups were facilitated at all sites by Martha Terry, PhD, of the University of Pittsburgh and lasted an hour. Food and nonalcoholic beverages were provided. All focus groups were audiotaped and were then transcribed. Analysis of common themes of discussion is underway and will be reported separately.
The VA has an advanced centralized electronic administrative records system in Austin, Texas, that allows for national tracking and information on all veterans receiving care from VA services throughout the country; this database also provides health care utilization information and ICD-9 codes.37,38
In addition, all VA medical centers have a Computerized Patient Record System that contains abundant demographic and clinical information useful to the study; in particular, the Computerized Patient Record System is used to extract laboratory data at each site with support from each local Information Resource Management office.39
The Pharmacy Benefits Management database provides medication information including type of medication, dose, and fill dates.38
As mentioned previously, BIRLS provides information on application for death benefits for VA patients.5,40
Data from BIRLS and inhospital mortality will complement data from the Cause of Death forms and will help to complete uncollected death data at the individual sites.
Each site has a team consisting of a Principal Investigator (an infectious disease provider), a Co-Principal Investigator (a general medicine provider), 1 or 2 study coordinators, and a member of the Community Advisory Board (CAB). Local IRB approval has been granted at each site.
The Coordinating Center in West Haven, Connecticut, serves as a source for survey design, development and printing, data analysis, and procedural protocol. Regular contact with sites is maintained through conference calls (PIs, study coordinators, writing committees, survey committees, data analysts) daily e-mails, telephone calls, faxes, and writing. In addition, the Coordinating Center drafts the Manual of Operations
, maintains the study website, receives completed surveys and electronic downloads, reviews proposals for data analysis and manuscripts, oversees the grant and budget, organizes scientific meetings, addresses problems at the sites, monitors recruitment, and keeps records of all IRB documentation including initial submissions, renewals, amendments, modifications and consent forms. The Project Coordinator makes annual visits to each site to ensure standardized procedures across all sites, review IRB documentation, observe consent and recruitment procedures, organize focus groups, check patient files for accuracy and protection from unauthorized access, and meet with all members of the site study team to address problems. In addition, an annual face-to-face scientific meeting is held near one of the participating sites. Details of the administrative structure of VACS can be found on the study website: www.vacohort.org
In an effort to describe the landscape of alcohol-related effects in chronic HIV infection and to contrast them with uninfected comparators, we have collected a wide range of alcohol and substance use measures (, see editorial in this issue). These have included measures of screening, diagnosis, consequences, dependency, expectancies, and symptoms. We conducted telephone interviews to collect current and lifetime patterns of consumption on half the sample enrolled at baseline and a quarter of the sample enrolled at the first 12 month follow-up. Some instruments, such as the stages of change and readiness to change for alcohol and cigarettes were collected expressly with the intent of developing behavioral interventions.
Data Sources and Analysis
To give the most up to date summary of VACS, we report enrollment as of September 30, 2004 (n = 5998). VA data reported in relation to target enrollments, participants versus nonparticipants, and the National VA sample come from administrative data (age, race/ethnicity, and gender), the VA EMR (pharmacy and laboratory packages), and all available sources (listed above). Centers for Disease Control data represent persons living with HIV/AIDS from the 33 states reporting in 2003.
HAART was defined, after consultation with our clinical experts, as being on at least 3 antiretrovirals at once. To make groups comparable with respect to proportion on HAART () regimens reported we restricted all samples to those for whom: first date of HIV care occurred after fiscal year 1998, who had a CD4 cell count from 30 days before first HIV diagnosis to 7 days after, and who had at least some pharmacy use. The sample sizes for these proportions are restricted with respect to the overall sample for each source reported (VACS n = 660/2979; nonparticipants n = 432/2151; virtual cohort n = 5059/33420). All p values reported in were estimated using binomial for proportions and difference between 2 population means for continuous data.
HIV Infected: VACS Participants, Nonparticipants, and VA and CDC National Benchmarks
Data reported in and come from the patient self completed survey. Of note, these data vary from administrative data on race/ethnicity (reported in ). Administrative data on race/ethnicity is determined by inspection by the clinic clerk whereas the survey data reflects the patient’s own determination. While patient self report is preferred, targets could only be established using administrative sources. P values reported in reflect results of binomial tests and χ2 tests.
Health Behaviors and Conditions