Sharing Surveillance Data across Jurisdictions: The DC/MD/VA Model Anne Rhodes, Virginia Colin Flynn, Maryland Rupali Doshi, District of Columbia Marcia Pearl, Maryland 1
Outline • Background • Black Box technology • Black Box iterations/ results • S TD data sharing • Future Directions 2
National HIV Care Continuum DATA and PROGRAM 3
Data Quality: What, Why, How? urveillance, Ryan White, and other HIV data are • S not j ust utilized for funding oddslot formulas and static reports • Real-time tracking of diagnosis, linkage, care engagement, medication adherence and viral suppression are needed • Current data systems – set up artificially with barriers based on funding streams, j urisdictions, disease status, etc. 4
Accuracy How do people get included in/ excluded from Continuum of Care analyses? • Death • Proof of out of j urisdiction address • No care in xx period of time? • Modeling methods? • Only care in xx period of time? 24% of current living cases in VA HIV S urveillance system – no lab in last 5 years (n=6,005) 5
Completeness • Markers for care cannot all be tracked in current HIV S urveillance system • S ystems outside of health department purview often have data on care status for PLWH • Electronic medical records/ health information exchanges/ all payer claims databases often available in j urisdictions 6
Timeliness – 4 th Goal calls to “ strengthen the timely • NHAS availability and use of data” • National viral suppression rates for 2013 for persons diagnosed with HIV as of 12/ 31/ 2012 (and alive as of 12/ 31/ 2013) released in July 2016 • AIDS .GOV site has care continuum with 2011 data 7
Black Box: Real Time HIV S urveillance • Pilot proj ect from Georgetown, funded by NIH. Involved DC, MD, and VA Departments of Health • RIDR de-duplication proj ect, funded by CDC. Used data from 8 j urisdictions: DC, MD, VA, NYS , NYC, WV, DE, NC, FL • Utilizes privacy technology for sharing surveillance data among j urisdictions where an algorithm for matching was set up in the “ black box” and returned matches of varying strengths (Exact to Very Low) to each j urisdiction 8
9 ATra™: A new methodology for co-analyzing non-shareable data Policy Body Pattern Matches Patterns Organization 1 Organization 2 Organization 3 Organization n 9
Regional HIV data sharing 10
S equence of Events in Cross- Jurisdictional HIV Data S haring 2013 2014 2015 2016 2017 • Dat a sharing • Dat a sharing • Black Box pilot • RIDR proj ect • eHARS dat a agreement s – agreement s for DC/ MD/ VA begins wit h 8-10 exchanges begin discussions begin signed complet ed j urisdict ions (large file back t o 2015), followed by • Weekly calls prospect ive files wit h j urisdict ions for DC/ MD/ VA 11
Matching - Pilot HIV S urveillance Records: 1981-2015 • Tot al (N=161,343) • Dist rict of Columbia (N=49,326) • Maryland (N=66,200) • Virginia (N=45,817) Mat ching Variables: • Last name of HIV case; • First name of HIV case; • Dat e of birt h of HIV case; • S ocial S ecurit y number of HIV case; • Hierarchical race/ et hnicit y assignment for HIV case; and • Last name soundex of HIV case 12
Matching - Pilot 13
Black Box Results - Pilot Output of person-matching across DC, MD, and VA eHARS databases: Over half of matches were not known to jurisdictions 14 14
Data S haring • Exchanged data files with identification variables after Black Box match and each state validated the accuracy of the matches • Over 90% acceptance for high, very high, and exact matches • Exchanged data files on accepted matches with data on diagnoses, demographics, risk, lab tests, residence, and vital status • Used to update records, improve data quality, and generate new HIV care continuum 15
Black Box Results for VA (RIDR): August 2017 Match, Exact and High Categories 100% 90% 80% 302 679 70% 159 3056 2626 11405 1752 1278 1553 60% 50% 40% 30% 294 491 20% 104 1902 1513 6417 856 629 628 10% 0% DC MD NC FL NYC NY st ate WV DE Tot al Not Known Known 36% of matches in exact and high categories not previously known to Surveillance program 16 16
S TD Data S haring Cross-j urisdictional case investigations – index cases and named partners Monthly conference call Maryland S ecure FTP site Department of Health High volume DC Department clinical site of Health (LGBT focus) Virginia Department of Health 17
Cross-Jurisdictional Case Investigations Letter to Providers, January 13, 2017 S igned by HIV/ S TD Leadership of DC, MD and VA Departments of Health “ … Currently, the three health departments actively cooperate and share information on persons who seek medical care outside of their area of residence. We must operate in this way to prevent new infections and assure individuals are linked to and retained in care and treatment. Please j oin us in in this cross-j urisdictional effort to increase the timeliness and effectiveness of our public health efforts to intervene in the spread of HIV and S TIs. As a front-line health care provider, you and your office staff have access to critically important information that can aid the health departments in responding to new HIV and S TI cases. Therefore, on behalf of each of our health departments, we authorize and encourage you to respond to requests for information on HIV and S TI disease investigations of cross-j urisdiction cases from our partner health departments in the National Capital Region… ” 18
Results: S o Far Increased Number of Improved Accuracy of Care Markers for Case Numbers Continuum • After address and • Care Markers vital status outside of eHARS , updates, number of added 8% to PLWH living in retention rates in Virginia as of 2014 and 9% to viral 12/ 31/ 2015 was suppression rates in reduced by 760 2015 persons 19 19
Results: Continued • Ongoing data sharing in DC/ MD/ VA – monthly meetings to discuss issues • Proj ects across j urisdictions, including Data to Care, cluster investigations and coordination of prevention and care efforts • Improved communication among j urisdictions 20
Future Directions Continued Building Quarterly Relationships Data S haring S haring National across Data S haring diseases 21
Challenges Leadership understanding and buy-in Bandwidth Comfort level with sharing identified disease data Information technology to support the proj ect Technical expertise Proj ect management 22
Final Thoughts • Data Improvement strategies should be part of plan for addressing • S haring data across j urisdictions is important for tracking disease and care for S TDs and HIV • Utilizing data for public health impact requires merging of multiple sources of information across systems, agencies, and funding streams 23
Acknowledgements CDC: Benj amin Laffoon, Dr. Irene Hall DC Department of Health: Michael Kharfen, Garret Lum, Auntre Hamp, Adam Allston, Brittani S aafir-Callaway, Toni Flemming, Deontrinese Henderson, Alberta Roye, Francoise Uwimana Georgetown University: Jeff Collman, Joanne Michelle Ocampo, J S mart, Raghu Pemmaraj u HRS A: Jessica Xavier, John Hannay Maryland Department of Health: Colin Flynn, Reshma Bhattacharj ee Virginia Department of Health: Lauren Yerkes, Kate Gilmore, S ahithi Boggavarapu, S onam Patel, Amanda S aia 24
Recommend
More recommend
