Bridging the Gap: Conference Record [Abstract book, International Conference on AIDS (12th: 1998: Geneva, Switzerland)]

928 Abstracts 43420-43423

12th World AIDS Conference

monthly basis. Finally, in late 1996 the database was directly linked to the CD4
Surveillance database, which had been operating since 1992. For the purposes
of confidentiality, soundex coding of name is extensively used by SCIEH on all
attributable HIV surveillance systems.
Results: As at 31 December 1997, 2725 HIV-positive reports and 899 AIDS
registrations had been received by SCIEH. Of the 2725 cases, 1011 are known
to have died. 1254 of the remaining 1714 are presently undergoing CD4 monitoring, 131 of these with a most recent CD4 count of <50, 291 between 51-200,
585 between 201-500, and 247 with a count of >500. AIDS registrations have
been received for 186 cases currently undergoing CD4 monitoring. Evidence of
previous negative testing has been traced in a total of 207 cases.
Conclusions: The integrated dataset (HIV Neg, HIV Pos, CD4 Count, AIDS,
Death) provides Scotland with an extremely powerful, and perhaps unique, tool
capable of monitoring national and local trends in infection as well as an individual's movement through different stages of HIV disease progression. The
database has already proved to be successful in providing detailed information
for revenue allocation and predictions of cases with severe HIV-related disease.
SCIEH hopes to add data on viral load testing at a future date.
S43420 Respondent-driven sampling: a new approach to
sampling hidden populations
Douglas Heckathorn, R.S. Broadhead'. Dept. of Sociology, 'Univ. of
Connecticut, Storrs, CT, USA
Background: Many persons at risk of HIV infection are members of "hidden
populations" because no sampling frame exists and public acknowledgment
of membership in the population would be potentially threatening. Standard
probability sampling methods inapplicable. Therefore, sampling often employs
snowball-type methods. Because the initial subjects from which sampling begins
(the "seeds") can almost never be drawn randomly, these are generally viewed
as convenience samples.
Objective: Drawing on a combination of Markov-chain and biased-network
theories, we predict that the sample will reach an equilibrium that is independent
of the initial subjects from which it began. We also predict that this equilibrium
will be reached at a rapid (geometric) rate. Hence, if chain referrals continue
through a sufficient number of waves, the bias introduced through the choice of
initial subjects can be eliminated. If so, the choice of initial subjects need not be
a source of bias. Our objective is to empirically evaluate these predictions.
Method: A sample of 277 active injection drug users was drawn from small
cities in central Connecticut using Respondent-Driven Sampling, method that produces unusually long (i.e., more than 12 stage) referral chains. Analyses focused
on the representativeness of the sample.
Results: Strong homophily - recruiting persons like oneself - were found with
respect to race/ethnicity. Homophily was weak for drug preference (opiates versus
cocaine) and nonexistent for age and gender. Despite the presence of homophily,
the sample reached equilibrium as predicted, and this required less than seven
stages. Therefore, the bias introduced by the selection of initial subjects had been
eliminated.
Conclusions: Chain-referral samples need not be dismissed merely as convenience samples. If referral-chains can be made sufficiently long, and thereby
penetrate sufficiently deeply into the networks of a hidden population, they can
draw nonbiased samples of the population.
Supported by the national institute on drug abuse R01 08014.
43421 Assessment of the impact of a major HIV/AIDS
confidentiality breach, Florida
Thomas Liberti12, L.A. Conti2, S. Lieb2, S. Dulin2, L. Maddox2, L.K. Crockett2.
Florida Department of Health 1317 Winewoood Bvd Tallahassee FL
32399-0700; 2Bureau of HIV/AIDS, Fla. Dept. of Health Tallahassee FL, USA
Background: In September 1996, a computer diskette containing the names and
demographic information of numerous HIV-positive patients was anonymously
mailed to a local health department and 2 major newspapers serving the 5-county
Tampa Bay area of Florida. The newspapers publicized the receipt of the unauthorized information, but did not further disclose the identities of the patients.
Investigations of the incident were launched by the Florida Department of Health
(DOH) and the Florida Department of Law Enforcement.
Methods: Law enforcement officials gathered evidence about the incident.
The DOH conducted a study to assess the impact of the breach on the AIDS
surveillance system and HIV counseling and testing programs.
Results: A DOH employee and a non-employee were implicated in the confidentiality breach. The volume of AIDS case reports declined temporarily in the
county where the employee worked due to his being placed on administrative
leave and the time consumed by the criminal and internal DOH investigations.
There was no evidence to suggest that local provider confidence in the AIDS
reporting system was adversely affected by the breach. When active AIDS surveillance resumed in early 1997, case reporting in the area returned to normal levels,
and missed cases were collected retrospectively. Elsewhere in the state, no effect

on case reporting was evident. Statewide, as well as in the 5-county Tampa Bay
area, overall HIV counseling and testing patterns for October-December 1996
were similar to those for the same period in 1995. In the county where the
employee worked, a slight decline in HIV testing was observed in the immediate
post-breach period, which was followed by a return to the pre-breach level. Implementation of legislatively mandated HIV infection reporting - initially scheduled
for January 1997 - was postponed by 6 months.

Conclusions: The breach had minimal impact on the AIDS surveillance system
and HIV counseling and testing programs statewide. The employee resigned, and
he and the non-employee were convicted of criminal charges. No class action or
individual lawsuit was filed by patients. Following the breach, DOH security and
confidentiality measures were enhanced statewide, and such measures became
a model for other states' HIV/AIDS surveillance systems. HIV infection reporting
was successfully implemented in July 1997.
43422 Surveillance of HIV-1 among young Thai men, 1990-1997
Suchai Kitsiripornchai1, L.E Markowitz2, P. Chanbancherd3, N. Limpairojn3,
N. Sirisopana2, S. Shou4, A. Brown5, A. Jugsudee3. 'AFRIMS 315/6 Rajvithi
Rd.; 2USAMC-AFRIMS; 3Army Institute of Pathology; 4RTAMC-AFRIMS;
5HMJF/USAMC-AFRIMS, Bangkok, Thailand
Objectives: To evaluate regional and temporal changes in HIV-1 prevalence, and
demographic correlates of HIV seropositivity among young Thai men entering the
Royal Thai Army (RTA).
Methods: Each year about 50000 young Thai men are conscripted into the
RTA. Since 1989, all conscripts have been screened for HIV-1 antibody; since
1991 limited demographic information has also been collected. Regional seroprevalence was based on region of residence before induction. Annual percent
change in prevalence was calculated overall and for each region (Upper North,
Lower North, Central, Northeast, South and Bangkok). Percent change, for each
region, from the year of highest prevalence was also determined.
Results:

Year
Prevalence (%)
Annual % change

1989 1990 1991 1992 1993 1994 1995 1996 1997
0.5 1.9 2.9 3.5 3.6 3.0 2.5 2.1 1.9
- 280 53 21 3 -17 --17 -16 10

In 1994, overall prevalence decreased for the first time to 3.0%, followed by
annual decreases through 1997. Since 1993, prevalence has decreased in all
regions except the South. By 1997, the prevalence in the upper North (3.8%) had
decreased 70% from its peak in 1992 (12.5%). The prevalence in South (2.6%)
was twice that in 1991 (1.3%). In all years, 21 year-olds had lower prevalence
than did 22-29 year-olds. Unmarried men had higher prevalence than did married
men from 1991-1993, but not after 1993.
Conclusions: The HIV-1 prevalence in young Thai men decreased 47% in
1997 compared to 1993. The overall percent change in seroprevalence has been
stable for the last 4 years. However, trends in prevalence varied by region. Reasons for the increase in prevalence in the South should be investigated; more
attention should be focused on this region.
143423 Toward an integrated surveillance system for HIV/STD
risk behaviors
Ronald R. Fichtner, C.A. Rietmeijer, R.M. Brackbill. Behavioral Surv Workgroup,
CDC-NCHSTP/OD Mailstop E08, 1600 Clifton Road, Atlanta 30333, USA
Issue: The development and evaluation of HIV/STD prevention programs are
increasingly guided by behavior rather than by disease indicators. While a variety
of surveys and surveillance systems yield HIV/STD-specific behavior information,
the utility of these systems for the development of locally-relevant interventions
may be limited due to lack of standardization, a scarcity of information on populations at highest risk, and limited efforts to synthesize and return data for local
use.
Project: In the fall of 1997, the US Centers for Disease Control and Prevention
(CDC) initiated a broad collaboration to improve the availability, quality, and use
of HIV/STD-specific behavioral information at the local level. Objectives include:
1) to identify (prototypes of) existing surveys/surveillance systems with greatest
local relevance; 2) to develop modules of standardized behavioral questions; and
3) to develop a system for data synthesis and dissemination.
Results: 1) Surveys/surveillance systems were identified at three levels: a)
general population (e.g., the Behavioral Risk Factor Surveillance System, the
Youth Risk Behavior Survey, and a variety of periodic or onetime surveys);
b) infected populations (e.g., HIV/AIDS Reporting System, Supplement to HIV
and AIDS Surveillance, STD surveillance/MIS, hepatitis surveillance); c) high-risk populations (e.g., the multi-site HIV Testing Survey, various local surveys.
2) Standardization will be achieved by the development of a hierarchical set
of behavioral questions for inclusion across survey instruments, with the "top"
questions being implemented at all levels and more specific items at the level
of infected and high-risk populations. 3) The ultimate outcome of the initiative
is to synthesize common data elements from the different survey levels. As an
example, a comparison of different surveys that already contain a common data
element (i.e., condom use at last sex) is being submitted as a separate abstract
to this conference.
Lessons Learned: Locally-relevant behavioral information is needed for the
development of HIV/STD prevention programs. The usefulness of data at every
level may be greatly improved by standardization of data items and by enhancing access of this information to planning groups. In addition, this process may
spur the development of local surveys. As methodological issues are not limited
to CDC-supported systems, we invite survey researchers and other interested
parties, both nationally and internationally, to participate in this process.