The ALPHA network.

Slides:



Advertisements
Similar presentations
Absorption, Retention and Empowerment
Advertisements

Antiretroviral therapy eligibility at enrollment and time to treatment initiation in Ethiopia Chloe A. Teasdale 1, Chunhui Wang 1, Sileshi Lulseged 1,
Development and pilot an automated Pregnancy and Birth Registry Kara Wools-Kaloustian M.D. M.S.
Unit 4: Monitoring Data Quality For HIV Case Surveillance Systems #6-0-1.
Surveillance to measure impact of ART Theresa Diaz, MD MPH CDC Global AIDS Program.
A Valuable Resource: Health Sector as a Beneficiary and Contributor to CRVS Systems.
HIV research in the era of ART: changing priorities in Tanzania Basia Zaba SOAS 3 rd March 2011.
Unit 1: Overview of HIV/AIDS Case Reporting #6-0-1.
Tracking of Inter-Facility Patient Transfers and Retention on Antiretroviral Treatment in Namibia Presenter Naita Nashilongo Ministry of Health and Social.
Attrition in HIV Care Attrition in HIV Care: Key Operational Challenge in implementing HIV Care and Treatment in Tanzania G R Somi _________ Ministry of.
eHARS to CAREWare Pilot Project Update and Training
Components of HIV/AIDS Case Surveillance: Case Report Forms and Sources.
Orientation on HIV care and ART Recording and Reporting System.
Matthew Fox Center for Global Health & Development Department of Epidemiology Boston University Health Economics and Epidemiology Research Office July.
Factors Associated with Survival in HIV-Infected African Patients on Antiretroviral Therapy: The Impact of a Sampling-Based Approach to Address Losses.
From Mekong to Bali: The scale up of TB/HIV collaborative activities in Asia- Pacific, August 8-9, 2009 Denpasar, Bali, Indonesia "TB/HIV Monitoring and.
HIV Care Continuum New Diagnoses, 2011, Fulton County, Georgia.
HIV Care Continuum Persons Living With HIV, Georgia, 2012.
Understanding temporal trends in HIV prevalence, incidence and ARV Dr Valerie Delpech Head of HIV surveillance Public Health England.
HIV and STI Department, Health Protection Agency - Colindale HIV and AIDS Reporting System The threshold for an ART secondary prevention effect on HIV.
Strengthening Cause-of-death Information in countries through Africa Programme on Accelerated Improvement of Civil Registration and Vital Statistics System.
A new method for estimating national and regional ART need Basia Zaba, Raphael Isingo, Alison Wringe, Milly Marston, and Mark Urassa TAZAMA / NACP seminar.
Medical Certification on Cause of Death Session V: Verbal Autopsy.
HIV Care Continuum New Diagnoses, 2011, Georgia. Persons with HIV Engaged in Selected Stages of the Continuum of Care, United States Percent
CD4 trajectory among HIV positive patients receiving HAART in a large East African HIV care centre Agnes N. Kiragga 1, Beverly Musick 2 Ronald Bosch, Ann.
From Aggregate Indicators to Impacting Patients - Data Use to Inform Treatment and Improve Care Ian Wanyeki Track 1.0 Implementers Meeting Dar Es Salaam.
Improving health worldwide Implications for Monitoring of the HIV Care Cascade? Jim Todd MeSH Satellite Session IAS Durban, Monday 18 th.
Beyond Counting – Using HIV Surveillance Data to Monitor Linkage to Care Following Release from Corrections Liza Solomon DrPH, MHS 9 th Academic and Health.
Boston University Slideshow Title Goes Here District Prevalence of Unsuppressed HIV in South African Women: Monitoring Programme Performance and Progress.
The ALPHA network. WHAT MAKES ALPHA DATA UNIQUE? An introduction to the network and its data resources Basia Żaba ALPHA network PI.
Boston University Slideshow Title Goes Here Eliminating CD4 thresholds in South Africa will not lead to large increases in persons receiving ART without.
Priscilla Tsondai, Lynne Wilkinson, Anna Grimsrud, Angelina Trivino,
New WHO Guidelines on Person centred monitoring
Factors associated with loss to follow up in a primary healthcare clinic practicing test and treat Authors: Julius Kiwanuka1,2, Noah Kiwanuka3, Flavia.
HPTN 071 (PopART): Have we reached the targets after two years of the PopART intervention IAS Paris July 2017 Richard Hayes.
Dr. Kęstutis Adamonis, Dr. Romanas Zykus,
Differentiated Monitoring & Evaluation
Men are absent across the HIV continuum of care in a rural area of southern Mozambique Laura Fuente-Soro, Elisa Lopez-Varela, Orvalho Augusto , Charfudin.
Contribution of ALPHA results to global estimates and policy
Earlier treatment and lower mortality in infants Initiating ART at
Acceptability of early HIV treatment among South Africa women N Garrett, E Norman, V Asari, N Naicker, N Majola, K Leask, Q Abdool Karim and SS Abdool.
Entry into care Failure to initiate timely HIV care after diagnosis is common ~75% of newly diagnosed link to care within 6-12 months Delayed entry into.
Illustrating the HIV Care Continuum in U.S. Cities
MeSH: Optimizing the use of routine HIV data
NYSDOH AIDS Institute Quality of Care Program eHIVQUAL
L.F. Jefferys1, J. Hector1, M.A. Hobbins2, J. Ehmer2, N. Anderegg3
David Culliford, Lynn Josephs, Matthew Johnson, Mike Thomas
A COLLABORATIVE APPROACH TO ESTABLISH PREDICTORS
HIV Diagnosis and the Cascade of Care in Ontario
Monitoring the implementation of the TB Action Plan for the WHO European Region, 2016–2020 EU/EEA situation in 2016 ECDC Tuberculosis Programme European.
Dr. Velephi Okello, Principal Investigator, MaxART Trial
Community patient tracking by Lay Community Health Workers (CHWs) is an effective strategy towards the 2nd & 3rd 90 Morapedi Boitumelo M.
Knowing your epidemic and knowing your response – maximising routinely collected data to measure and monitor HIV epidemics in sub-Saharan Africa Monitoring.
Sources of vital statistics
Disclosures: no disclosures
From toward HIV Elimination with Boosted-Integrated Active HIV Case Management (B-IACM) in Cambodia Dr. Penh Sun LY, Director, NCHADS Presented.
Talent Maphosa; MD, MDS, Dip HIV Man Technical Director, OPHID
Retention: What It Means for You
Management and Development for Health (MDH)
Poster WP41; Contact: David A. Katz,
Illustrating the HIV Care Continuum in U.S. Cities
Surveillance of Tuberculosis
monitoring & evaluation THD Unit, Stop TB department WHO Geneva
Epidemiological Measurements of health
Illustrative Cluster Detection and Response Strategy
Stakeholder engagement and research utilization: Insights from Namibia
Knowledge of HIV Status in Kenya
Why Quality Matters in ART Programs
Treatment Outcome among patients on ART in Southern Tanzania: Does Time of ART initiation Matter?
HUMAN IMMUNODEFICIENCY VIRUS (HIV) PREVENTION & CARE
Presentation transcript:

The ALPHA network

Verbal Autopsy ALPHA Network Contributions Data: ALPHA Network Collaborating Institutions Methods Development & Results: Zehang (Richard) Li (UW) Clara Calvert (LSHTM) Tyler McCormick (UW) Basia Zaba (LSHTM) Samuel Clark (UW) Helpful Discussion: Jon Wakefield (UW) Peter Byass (Umea)

Verbal Autopsy VA can assign causes to deaths when certified medical autopsy not possible VA is comparatively affordable and feasible VA-derived causes are less accurate than medical autopsy ALPHA is working to improve VA in general and for populations with HIV

Vital Statistics Performance Index (range 0 to 1) Why VA? Vital Statistics Performance Index (range 0 to 1) Source: Lene Mikkelsen et al. 2015. “A global assessment of civil registration and vital statistics systems: monitoring data quality and progress”. Lancet 386: 1395–406.

ALPHA Network VA Traditional VA not very good at identifying HIV-related deaths ALPHA is addressing this Adding new questions to VA interviews Linking deaths to clinic records Developing new automated cause assignment methods Using known HIV status to evaluate performance of VA for HIV

Physician Review Trained physicians read VA interview transcript and assign causes Because physicians sometimes disagree, standard approach is 2+ physicians review each VA Key limitation: physician bias Inefficient use of physician time / expensive Long delays in reading VAs

Automated Assignment Standardizes cause assignment Limitations Cheap Reproducible Comparable Limitations Less specific / precise than physicians Incorporates less information No clear standard approach

InSilicoVA1 Builds on existing method InterVA2 Based on physician-derived information relating VA symptoms to causes Adds ability to use information on symptom-cause relationship from de-biased physician-assigned causes Ensures individual causes and distribution of causes are consistent Quantifies uncertainty 1 https://arxiv.org/abs/1411.3042 2 www.interva.net

HIV-related Deaths

Physician Information Add de-biased information from small fraction of deaths coded by physicians

WHO Standards ALPHA VA team participates in WHO working group on VA Standardizing VA for use in both research and routine surveillance settings WHO ‘2016 Standard VA Instrument’ Supports all software: InterVA, Tariff, InSilicoVA Electronic instruments (ODK) Manual and training materials Coming …

ALPHA VA Software ALPHA VA Team has developed free, open-source software to run all reasonable automated methods Implemented in free, open-source, multi-platform statistical package R Both R and ALPHA software ‘packages’ available for download on CRAN - Comprehensive R Archive Network: https://cran.r-project.org

ALPHA R Packages InSilicoVA (ALPHA) InterVA4 (Peter Byass) Tariff (IHME) openVA Runs all three Data conversion Visualizations Comparison of results

Improvements Standardization of VA interview across linguistic and cultural settings Full use of VA narrative section in automated methods Better and more diverse training data for automated methods, archive of VA with: De-biased physician-assigned causes MITS-assigned causes Autopsy-assigned causes

Improvements Statistical methods development Continued contribution to and coordination with WHO VA working group ‘Slimmed down’ VA for use in vital statistics and routine surveillance Work with other disease-specific research applications, e.g. Ebola

Linking research to routine service data National AIDS Control Programme, MoHCDGEC, Tanzania Collaborators: Dr Geoffrey Somi Joseph Nondi Werner Maokola Prosper Njau Renatus Kisendi Michael Mahande Jenny Renju Paul Mee Jim Todd

Rationale ALPHA data provide gold standard estimates in limited locations Population cohorts provide: Incidence HIV, access to services, impact (mortality) of HIV Linkage of population data to health facility data, provide estimate of access at each stage of the treatment cascade Routine clinic-based data can learn from the ALPHA analyses All facilities provide aggregate data for monitoring HIV indicators Increasing number of clinics have patient data (EMR) Benefits of using ALPHA analyses on patient data from clinics New WHO recommendations for case-based surveillance Organising patient-level data to estimate the treatment cascade Linkage of data between different clinics using unique ID Analysis programmes can be developed using ALPHA methods Alpha population cohorts provide gold standard estimates of the treatment cascade, including incidence of HIV and impact, or mortality associated with HIV. Linkage of population data to the clinics serving the population enables estimates of access to the treatment cascade However there are a large amount of data available from national programmes, and we can use some of the ALPHA methods to analyse these data. The data include the regular routine reporting of aggregate data for national indicators. An increasing number of clinics now have electronic patient data which can be analysed using the methods developed by ALPHA In the future the implementation of case-based surveillance will improve the data available for analysis, as it will make patient level data more comprehensive, and improve the unique ID of patients to allow linkage across clinics

CTC in Tanzania - 2014 Alpha cohorts Care and Treatment clinics (CTC) This map shows the health facilities in Tanzania, with the care and treatment clinics (CTC) shown as coloured boxes. The black dots are the health facilities that do not provide ART at present, but many of them are now part of the PMTCT programme. As you can see CTC is available across the whole country The 2 ALPHA cohorts in Tanzania provide detailed estimates of the treatment cascade in Ifakara and Kisesa. Using the data from the CTC clinics we can translate that into a national figure for mortality in Tanzania.

Coverage by age & sex Among those attending CTC in 2014: Females more likely to be on treatment. 50% of the patient are on ART for 2 years or more. These are 2014 data from the 650 clinics with electronic data analysed for the Tanzanian CTC report this year. These show percentages in each sex and age group, but there are twice as many females as males in the CTC database. Among those attending CTC females are more likely to be on treatment than males (which agrees with the Hazard ratios shown from the Alpha data). However these data have a greater proportion of patients that are on ART for 2 years or more reflecting the difficulty in retention.

ART initiation 2011-2014 450,000 ART naive patients at first visit in 637 clinics: 33% eligible for ART at enrolment at first visit 8.5% died before ART initiation (from sub-study). Mortality Rate of 68.4/1000 person years Nationally, between 2011 and 2014 for new patients Median time to ART initiation from 154 days in 2012. down to 40 days in 2014. Median CD4 count: 356 in 2011, 300 in 2014 This slide shows results from the analysis of the national data from 2011 to 2014, which is part of the 2016 CTC report. A third of those were eligible for Art at first visit. This is a big increase from the 8% that were eligible at first visit in 2008.

Mortality rates Mortality in Care & on ART (2010-2013) Similar to ALPHA No data prior to CTC enrolment Limitation for mortality – LTFU unknown For ART > 6months Excess mortality = >double HIV negative Estimates of mortality at different points in the treatment cascade can be obtained. From the CTC data we have no estimates of the mortality among the undiagnosed, or among those who have not yet attended the care and treatment clinic. At all stages of the cascade mortality among males is higher than mortality among females. The estimates of mortality among those not yet on ART, and in the first six months on ART are similar to those from the Alpha data. Mortality after six months on ART is lower than on the previous stages of the cascade, but is still more than double the background mortality among HIV negatives. The major limitation of these data is that we do not know what happens to those lost to follow up Data from 637 CTC across Tanzania – CTC report #4 (2016)

Retention over time From 2006 to 2010 – large increase in CTC enrolment 40% retained in same clinic after 3 years Of those retained 1 year after enrolment – 90% initiate ART Of those retained 3 years on ART – 80% successfully treated Retention is difficult to measure among patients who move clinics, as we do not have data on those who move to a clinic without the electronic database. Inn Tanzania there was a large increase in those enrolled at CTC between 2006 and 2010. Following those cohorts for three years we see less than 40% were still attending the same clinic three years later. But of those who were retained 90% initiate ART within one year, and 80% of those on ART had clinical treatment success (we dont have viral load data, but treatment success is based on clinical staging)

Discussion Advantages of routine clinic data Larger, representative numbers for analyses Accurate dates for enrolment and ART start Additional analyses – adherence, CD4 counts Limitations (as illustrated from ALPHA data) Data quality, including unique ID of patients Missing data on diagnosis, and death No end points for those LTFU Issues fed back to alpha for bias assessment Case-based surveillance advantageous Routine clinic data provides an immense resource for analysis which would be representative of the whole country. It has accurate data on the data of first visit to the clinic and the date of ART initiation, so does not rely on self reports. We have data on CD4 counts and adherence. However there are data quality issues, although these are less with greater use, and understanding of what the data can show. It is difficult to trace patients across clinics as we to strengthen the unique ID of the patients. And we do not have good data on patients who do not come to the clinics. The future is to feedback these results to the Alpha cohorts to establish the biases in these clinic estimates. We also look forward to the new investment in case based surveillance