Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Ethical Challenges Using Internet and Mobile Technology for HIV
Prevention Research
Brian Mustanski, Ph.D.
www.impactprogram.org
Overview
• My experience • Why use the Internet for HIV and other health
research? • What are the risks of everyday internet use
and how to people approach privacy? • Case studies • Tension of privacy versus data quality
www.impactprogram.org
My experience with online sexual health research 1999 LGBT College Student Survey of
how “having sex” is defined Recruited through LGBT University Student
Organizations. Online survey.
2002 Online daily diary study Recruited through free banner ads on high volume gay websites. Online survey and diaries.
2007 Project Q2 Recruited using respondent driven sampling (RDS). Follow up surveys could be done online.
2007 1.0 Recruited when testing HIV negative at clinics in Chicago. Online intervention RCT.
2011 Keep it up! 1.5 Recruited through Chicago community organization. Online intervention service.
2012 Recruited online with Facebook ads. Online sexual health education.
2012 Recruited online through banner ads. Evaluation of rapid at home HIV testing
2012 Recruited when testing HIV negative at clinics in Atlanta, Chicago, and New York. Online
intervention RCT.
2012 Guy to Guy Recruited online through LGBT youth oriented websites. Focus groups and text intervention RCT.
www.impactprogram.org
Share your experiences
• With online data collection • How did you go about detecting or preventing
fraudulent participation?
www.impactprogram.org
Why use the internet/technology for HIV and other health research
• High rates of use. • Some aspects of data collection are facilitated
– No need for data entry (still an issue offline?) – 24/7 data collection
• Access to geographically diverse populations • Assess to “hard to reach” populations
(sometimes through direct advertising on “mainstream sites” and sometimes through specific websites).
www.impactprogram.org
Internet use
www.impactprogram.org
Cell phone use
www.impactprogram.org
Use of technology for health purposes
• 80% of internet users, or 59% of U.S. adults, look online for health information (Pew, 2012).
• 17% of cell phone owners, or 15% of adults, have used their phone to look up health or medical information. Young people, Latinos, and African Americans are significantly more likely than other groups to have mobile internet access and to conduct health searches on-the-go.
www.impactprogram.org
What are the risks of every day internet/technology use?
www.impactprogram.org
How do teens think about privacy online?
Teens don’t always have a good understanding about how their personal data is used: Middle Schooler: “Anyone who isn’t friends with me cannot see anything about my profile except my name and gender. I don’t believe that [Facebook] would do anything with my info.” High Schooler: “I don’t know if Facebook gives access to others. I hope not.” High School Boy: “I don’t think [Facebook] should give anyone access to profile information.” High School Girl: “It depends on what kind of profile information they’d share. If it was only my age and gender, I wouldn’t mind. If they went into detail and shared personal things, I would mind!” High school boy: “I don’t think it would be fair because it is my information and should not be shared with others, unless I decide to.”
www.impactprogram.org
Impactprogram.org: What our analytics show
Repository for IMPACT’s innovative research Sexual health resources for LGBT youth Original videos Weekly blogs for youth and researchers
www.impactprogram.org
• Keywords searched for • How they got to site • What network they are on (e.g. Comcast,
Northwestern, US Department of Justice, etc).
www.impactprogram.org
Case Study
www.impactprogram.org
Overview of Just/Us Study • Goal to reach African American and Latino youth with HIV
prevention education on Facebook. – These groups are at higher risk compared to white youth and less
likely to be engaged in HIV prevention research.
• Enrolled N = 1,588 in an RCT through community recruitment, newspapers ads, and online network recruitment. Ages 16-25 with waiver of parental permission.
• Intervention involved multiple updates each day about sexual health, plus a facebook page with more information.
• Outcomes include attitudes towards, norms for, and implementation of healthy sexual behavior.
www.impactprogram.org
Ethical Issues
www.impactprogram.org
Ethical Issues
www.impactprogram.org
Ethical Issues
www.impactprogram.org
Case Study 2: G2G online focus groups
www.impactprogram.org
Tension of confidentiality/ease of enrollment versus data quality
www.impactprogram.org
Why does data quality matter? • What are the implications of fraud in online research?
– Compromised validity of findings – Waste of resources
• Increasing adoption of web-based data collection in health research
means we need to be aware of these issues or risk general skepticism of this method.
• Important to make sure data collected through this modality are valid
and reliable, and that conclusions are accurate. – Duplication – Falsification – Validity and reliability of responses (not the major focus of this
presentation)
www.impactprogram.org
How common are these issues?
• Invalid entries occur commonly in web-based research with MSM. – Konstan and colleagues (2005) found that 11% of
entries in their sample (N=1,150) were duplicate entries from participants.
– Bowen and colleagues (2008) found that
approximately 1/3 of the 1,900 total submissions were multiple entries.
www.impactprogram.org
How common are these issues?
• In Bauermeister’s work with young adults (18-24)... – In a sample of single YMSM filling out a survey,
10% were fraudulent/duplicate cases. • iTunes gift card ($15)
– In a sample of emerging adults (N=3,447), 16%
were fraudulent/duplicate cases. • VISA e-card ($20-$75)
www.impactprogram.org
Bockting’s NICHD-Funded “All Gender Health” Study
• ~70% fraudulent cases • Prevention used
– Phone number verification
• Detection used – IP address lookup – Mailing address lookup – PayPal name in receipt – Inconsistencies in longitudinal responses
www.impactprogram.org
Suggested Best Practices
• To be published in a forthcoming commentary
www.impactprogram.org
Preventing falsified data… • Recruitment
– Eligibility screeners that don’t explicitly state eligibility criteria and include sham questions
– Limiting URL placement through robot.txt instructions – Monitoring ULR awareness (google alerts, regular linkto
searches) – Targeting ads to only those who meet some eligibility criteria – Multiple steps in enrollment procedure to allow chance to very
email address and potentially other contact information. Run in waiting period (don’t find out if eligible until after a set time).
– Restriction of one submission per IP address or computer (using cookies)
– Use of CAPTCHAS (Completely Automated Public Turing test to tell Computers and Humans Apart)
– Make payment require provision of identifying information (name, mailing address)
www.impactprogram.org
Detecting falsified data… • Automated Procedures
– IP match – Time to completion – Consistency of responses (i.e. age matches DOB) – Track referral URL – Creating statistical algorithms for matches – Must call from phone into system to confirm reported number
• Manual Procedures – Cross-checking entries from similar IPs – Reverse IP lookup (issues with server location;
http://www.networksolutions.com/whois/index.jsp) – Flagging cases – The “you have been flagged” email – Compare information against public sources like facebook
www.impactprogram.org
Next generation fraud detection • Algorithms that can predict users demographic
characteristics based on provided information.
• Identity verification systems that seek to establish a mapping from a person’s online identity to their real life identity (e.g., LexisNexis Identity Management). Participants must provide personal data online, to the identity verification firm. The firm then checks public and private databases for a match and prompts the user to answer questions based on the records. An identity "score" is calculated, and the identity of the online user is either given the "verified" status, or not, based on the score. E.g. “What is your mother’s maiden name?”
www.impactprogram.org
Points for discussion
• Tension between making enrollment easy to facilitate participation and needing to prevent fraud.
• Use of precious financial and staff resources for fraud detection and prevention.
• IRB consideration.
www.impactprogram.org
My experience with online sexual health research 1999 LGBT College Student Survey of
how “having sex” is defined Recruited through LGBT University Student
Organizations. Online survey.
2002 Online daily diary study Recruited through free banner ads on high volume gay websites. Online survey and diaries.
2007 Project Q2 Recruited using respondent driven sampling (RDS). Follow up surveys could be done online.
2007 1.0 Recruited when testing HIV negative at clinics in Chicago. Online intervention RCT.
2011 Keep it up! 1.5 Recruited through Chicago community organization. Online intervention service.
2012 Recruited online with Facebook ads. Online sexual health education.
2012 Recruited online through banner ads. Evaluation of rapid at home HIV testing
2012 Recruited when testing HIV negative at clinics in Atlanta, Chicago, and New York. Online
intervention RCT.
2012 Guy to Guy Recruited online through LGBT youth oriented websites. Focus groups and text intervention RCT.