560 likes | 689 Views
A new online survey module to measure sexual partnership timing With results from a focus group of MSM. Eli S Rosenberg, Patrick S Sullivan Sex::Tech February 26, 2010. Department of Epidemiology Rollins School of Public Health Emory University, Atlanta, GA.
E N D
A new online survey module to measure sexual partnership timing With results from a focus group of MSM Eli S Rosenberg, Patrick S Sullivan Sex::Tech February 26, 2010 Department of Epidemiology Rollins School of Public Health Emory University, Atlanta, GA
Background [1]: Concurrency by example You have unprotected sex with A, B, C A B C • Timeline of interventions? time • C is at risk from B and A • B is at risk from A • A is not at risk from others Serial monogamy Another situation: A B C time Concurrency • Now A also at risk from B and C.
Background [2]: Concurrency, defined Concurrency: “Overlapping sexual partnerships where sexual intercourse with one partner occurs between two acts of intercourse with another partner.” Consultation on Concurrent Sexual Partnerships UNAIDS. Nairobi, Kenya, 2009 A B C time ≈ A A B C
Background [3]: Concurrency as amplifier of transmission risk • Concurrency accelerates HIV/STI transmission • Modest changes in concurrency can greatly impact transmission • Common among heterosexual young adults: • Concurrent partner during most recent relationship: • ♂ 31%, ♀ 26% (1995 survey in Seattle) • More prevalent among blacks • Timeline of interventions?
Background [5]: Concurrency in MSM • Little information about MSM • NHBS MSM-2, San Francisco (Bohl et al 2009) • First study to specifically look at concurrency in MSM • Some evidence for more concurrency among black MSM • A 2-year cohort of 680 HIV- black and white MSM in Atlanta, evaluating HIV and STI incidence. • Timeline of interventions? • HIV/STI testing • Risk behavior assessment • Individual level • Partnership level (including timing / concurrency) • Community level • Need for CASI behavioral assessments
Challenge:Difficulties in Measuring Concurrency • Ideal = Sex diary of who and when • Concurrency can be inferred • Not practical for our study • Two common solutions: • List all partners within the previous X months and then ask: • 1. “Once you began sexual activity with Joe, with how many other people did you engage in sexual activity?”(direct, Manhart et al ‘02) • or • 2. Start / end dates for sex for each partner (indirect) • Poor agreement between methods 1 and 2 (Manhart et al ‘07) • Primarily due to date recall and order • Dates often at the month-only resolution = ambiguities • Need better and smarter methods if using CASI • Timeline of interventions?
Overcoming challenges [1]: A third option – interactive calendar Concurrency Need to clarify Pattern! Concurrency Need to clarify all 3 pairs
Overcoming challenges [2]: Intelligent CASI survey • Implemented all 3 concurrency methods in SurveyGizmo • SurveyGizmo • Usual question types • Fancy ones too: Tables of checkboxes, rating scales, etc. • Can dynamically change based on user response: • Survey text • Question behavior • Page navigation • Export data to SPSS, Excel, CSV • Accessibility options • Integration with Facebook, cell phones, etc. • Timeline of interventions? • Interactive controls • PHP scripting
Solution: Partnership timing module (country-themed bar in ATL)
Let’s be mischievous and put an end date before a start date …. First had sex on 1/1/2010
We are sent back to beginning of dates section and need to correct
Concurrent or serial? Need to clarify!
What if I stopped having sex with Buck456 in January 2010? Need to clarify!
What if I had sex with Joe in December 2009? Obvious concurrency = do not need to clarify! Need to clarify!
Results [1]: Pilot testing and focus group, Dec. 2009 • Venue- and street outreach-based recruitment in Atlanta • Eligibility criteria: • Black or white non-Hispanic MSM • Not currently in a mutually monog. relationship • Atlanta resident • ≥ 18 years • > 1 partner in previous 6 mo. • 13 MSM took survey, 10 took part in focus group • 10 white, 3 black • median age of 30 years • median 4 partners in the previous 6 mo.
Results [3]: Relevant focus group comments • Preference for calendar method • “The grid at the end with the calendar, that sorta recapped everything… it was easier to think like that… it was easier to recall than one person at a time, with a date“ • “… most of them I just remembered generally, but I did know who was before who, so I knew exactly that” • Acceptability of follow-up questions to resolve ambiguities • “I liked that at the end where it came back and said, … ‘sex with these 2 people in October. If so, did you stop with him before you started with him’. That was so much easier for me. “
Results [4]: Suggested changes • Place the partner calendar at the beginning • “I would collect all of the relative stuff [calendar] up-front like this … because in our heads we know who these people are” • Only ask for month + year, allowing date opt-in • “… it’s just hard for me to remember exact dates. I was pretty much estimating.” • “Do the month thing first and then say, ‘if you have anything more specific, that would help us’ ” • A number of other helpful suggestions for language and • format
Conclusions • Concurrency an important determinant of STI/HIV transmission • We designed an improved survey with partner calendar to overcome measurement limitations • SurveyGizmo + scripting language allows for online implementation • Pilot testing demonstrates acceptability and preference over extant methods • NHBS-MSM, cycle 1: November 2003 - April 2005 • 13,670 men interviewed • 15 MSAs • Venue-based time-space sampling • Eligibility criteria: • Male, age ≥ 18 years, current MSA resident, provide informed consent • Analysis criterion: • ≥ 1 male sex partner in previous year
Thank you! supported by NIH grant # 1R01MH085600
Background [4]: What about MSM? • Men who have sex with men (MSM) are the most heavily impacted group in the United States • In 2006, comprised: • 53% of all new infections, overall • 72% of all new infections, among men • Within MSM, heterogeneity of risk • NHBS, venue-attending MSM, 5 cities ’04 - ’05 (n = 1,767): • Timeline of interventions? • Individual risk factors do not explain disparity (Millett et al ’06, ‘07) • Concurrency and other network factors?
Overcoming challenges [3]: SurveyGizmo Overview • Timeline of interventions?
Background [1]: SurveyGizmo shots • Timeline of interventions?
Overcoming challenges [4]: SurveyGizmo Overview • Timeline of interventions?
Background [1]: SurveyGizmo shots • Timeline of interventions?
Background [1]: SurveyGizmo shots • Timeline of interventions?
Background [1]: SurveyGizmo shots • Timeline of interventions?
Overcoming challenges [5]: SurveyGizmo Overview Powerful PHP scripting language extends the ability of the survey to adapt, respond, and perform advanced calculations.