The best magazine
Methods
We conducted a cross-sectional analysis of data from routine risk assessments and HIV/STI testing at 3 different HIV testing service models (described below) during the period August 2013 and May 2014 inclusive, all located within 2.5 km of the central business district of Sydney. Sydney has an estimated population of 4.8 million residents with some of the most densely populated areas located around the central business district and also has the highest concentration of gay men. The community-based testing service models aimed to see GBM exclusively and the fast track clinical service model saw GBM and other risk populations.
Fast Track Clinical Service Model
The fast track clinical service model (hereafter referred to as Xpress) began operation at Sydney Sexual Health Centre, the largest sexual health clinic in NSW, in December 2010. Sydney Sexual Health Centre is located in the Sydney central business district and also in close proximity to the gay precinct of Darlinghurst. Patients are greeted by reception staff and instructed to complete a computer-assisted survey instrument (CASI) on touch screen computers. This model has been previously described and evaluated. In brief, patients are seen by a nurse (usually an enrolled nurse) who performed the HIV rapid test; the client self-collected samples for STI testing; and the nurse performed venepuncture for HIV, syphilis, and hepatitis serology. On-site counselors were available in case of a reactive HIV rapid test result.
Community-Based Organization Service Model
The fixed-site community-based organization testing service model (hereafter referred to as Community site) involved a collaboration between Sydney Sexual Health Centre and ACON, which is NSW's leading health promotion organization specializing in HIV and lesbian, gay, bisexual, transgender, and intersex health. The service operated out of a purpose built clinical space located on the ground floor of the ACON office building in Surry Hills. The building is close to a central railway station hub where all Sydney metropolitan and country rail lines intersect. Surry Hills borders Darlinghurst and is located in the gay precinct of Sydney. The model began operation on June 13, 2013. The Community site involved being greeted by a peer educator who instructed GBM on how to use the CASI; the peer educator performed the HIV rapid test; the client self-collected samples for STI testing; and the nurse collected venepuncture for HIV, syphilis, and hepatitis serology. On-call counselors were available in case of a reactive HIV rapid test result.
Time-Limited Shop Front Service Model
The time-limited shopfront community testing model (hereafter referred to as Shopfront) operated out of a shop front on Oxford Street in Darlinghurst, Sydney Australia, and also involved a collaboration between Sydney Sexual Health Centre and ACON. Oxford Street is a major thoroughfare for traffic and pedestrians and is located in the heart of one of the busiest gay precincts in Sydney with the highest population of gay men and also the largest number of general practices, hospitals, and sexual health clinics, which specialize in gay men's health. The Shopfront operated for 6 weeks during the Sydney Gay and Lesbian Mardi Gras festival, which is held each year, and culminates with a parade that traverses Oxford Street. The patient flow through the service and STI and HIV testing was the same as described for the Community site; however, GBM were required to walk 2 blocks up the street to a local Gay bar to use the toilets for self-collected specimens as there was no toilet onsite in the testing venue.
CASI Variables Collected
The CASI used in the various models collected the client's age, postcode of residence, country of birth, language spoken at home, gender, and number of sexual partners in past 3 and 12 months, HIV testing history, condom use in the last 3 months, history of injecting drug use and commercial sex work in the last 12 months, and hepatitis vaccination history.
HIV and STI Testing
Men seen in all 3 HIV testing models were offered the Trinity Biotech Uni-Gold HIV-1/2 Rapid Test ("Uni-Gold", Trinity Bioteck, Bray, CO Wicklow, Ireland.). Men attending the Community site and the Shopfront were offered a choice between the Uni-Gold and the OraQuick ADVANCED Rapid HIV-1/2 Antibody test (OraQuick OraSure Technologies Inc. Bethlehem, PA). As rapid tests offered at the services were not approved by the Therapeutic Goods Administration at the time of the study, access to HIV rapid testing was only available as part of clinical research trials being conducted by the Kirby Institute and The Albion Centre. Ethical approval was obtained; consent was obtained from GBM on their first test, and conventional laboratory HIV testing was conducted in parallel. Patients were offered full HIV/STI screening in accordance with Australian testing guidelines for GBM including gonorrhea, Chlamydia, and syphilis.
Patient Flow
We calculated the total number of patient visits and the total number of testing staff hours at each of the service models to calculate the ratio of patients seen per testing staff hour. Total testing staff hours did not include greeters, administration staff, or on-call counselor time. We compared average patients per testing staff hour at each of the testing models by pairwise Poisson analysis.
We also calculated the average patient journey time in minutes for all attendances during the study period but as this was summary data therefore could not be statistically analyzed.
Statistical Analyses
We classified GBM as being at high risk of HIV infection if they reported more than 5 partners in the past 3 months or more than 20 partners in the past 12 months consistent with the Australian testing guidelines for GBM. These guidelines also include any recent unprotected sex in their definition of high risk, but the CASI does not differentiate condomless sex between regular and casual partners where the risk of HIV varies; therefore, we did not include it in our definition. We used the Australian Bureau of Statistics categories to assign region of birth and the NSW government postcode definitions to assign patients into Local Health District boundaries.
Gonorrhea and Chlamydia cases were defined as a positive result from a specimen tested with the Cobas (Roche, Pleasanton, CA) Amplicor CT/NG test. Infectious syphilis cases were defined based on State case definitions and used both clinical and laboratory test results. We calculated the overall HIV and STI test positivity by dividing the number of positive tests in new clients, by the number of new clients tested, and calculated 95% confidence intervals (CIs). STI positivity at a specific site was calculated as the number of men who tested positive for a specific STI at that site divided by the total number of men who were tested for that STI at that site.
We compared the demographic risk behavior and HIV testing history of new GBM patients attending the 3 testing models using χ or Fisher exact test and Wilcoxon rank-sum tests. We compared the average patients per testing staff hour at each of the testing models by pairwise Poisson analysis. We used 95% CIs to assess significant differences in HIV/STI positivity across the models.
We calculated 2 testing outcomes: untested (the proportion of GBM who were untested ever) and infrequent testers (the proportion of GBM who previously tested ≥12 months ago excluding untested men).
We conducted 2 separate logistic regression analyses: one with untested as the primary outcome, with the comparison group being men who had tested previously, and the second with infrequent testing as the primary outcome with the comparison group being men who tested frequently. We included a range of other sociodemographic and behavioral factors in the univariate logistic regression to account for other known correlates of the primary outcomes, as described in the introduction. It is possible that a service model may reach a higher proportion of untested men simply because they reached a much younger client group than other services. Therefore, variables included in the bivariate models were service model, age group, country of birth, language spoken at home, area of residence, if bisexual (reported both male and female sexual partners in past 3 or 12 months), number of male and female partners in the last 3 and 12 months, and injecting drug use ever. Multivariate models considered all factors with P < 0.10 in initial analyses and used forward stepwise methods. Final multivariate models were created that included all statistically significant factors (P < 0.05). All analyses were conducted using STATA 13 statistical package.
Ethics
The South East Sydney Local Health District Human Research Ethics Committee approved the study (reference 08/223).
Source: ...