1. The reality of case based surveillance: information for action in concentrated and low prevalence countries
Susan Cowan, MD
Head of section for bloodborne and sexually transmitted diseases,
Department of infectious medicine epidemiology,
Statens Serum Institut,
Denmark

2. Disclosure and thanks I have no conflict of interest
I thank Valerie Delpech, PHE for lending me her slides so I didn’t have to invent the wheel
I thank Simba Takuva, National Institute for Communicable Diseases, NHLS, South Africa, for the use of a slide from a CROI presentation, February 2015*
* Engagement within HIV Care in South Africa: Implications for Prevention

3. Need for data
Prevalence data : (STI clinics, Antenatal, Blood donors etc)
Used to ”model the epidemic”
Accurate number of HIV and AIDS-deaths are critical
Incidence data: Rutine case reporting of new diagnoses
monitor HIV trends in a country and among certain populations
Incident cases per prevalent cases = transmission rate
Clinical and epidemiological data
Monitoring changes within, and disparities between risk populations to inform public health decisions
Monitoring quality of care indicators
Modelling
CD4 is Key to monitoring late diagnosis, incidence and undiagnosed fraction estimates, probable country of infection estimates
VL is key to monitoring link to care and retention in care

4. Example of using several data sources
Vital Registration Data (Statistics SA)
Population mid-year estimates (2012)
Key Outputs
1. Number living with HIV
National Household Survey (HSRC)
HIV prevalence estimates (2012)
2. Number linked to care
3. Number on ART
4. Number with suppression
Laboratory-based Surveillance Data (NHLS)
CD4 Counts : proxy for # linked to care
(n = 3.9 million)
Viral Loads : proxy for # on ART
(n = 2.4 million)
Study population and datasets - the NHLS Electronic Data Warehouse is a repository for all public sector laboratory measurements conducted in South Africa. Between and 2013, At least 18million CD4 tests had been conducted. 3,9 million during
In this project, we constructed a database of HIV positive patients using CD4 count and viral load results.
Since this is specimen based data and multiple tests can relate to the same individual, probabilistic record linkage techniques were used to create a single record per person per year.
Variables incorporated into the algorithm include name, surname, initials, date of birth, address and gender.
Assumptions
Linked to HIV care: any person with a CD4 count is presumed to have been linked to HIV care following HIV diagnosis.
On antiretroviral therapy (ART): a viral load measurement is used a proxy for a patient on ART. In accordance with South African HIV treatment guidelines, patients on ART have a viral load at 6 months into ART then at 12 months and yearly thereafter for monitoring and viral load measurements are not indicated for patients not on ART.
Estimating on ART in 2012: we used the dataset for viral load measurements ranging from 1 January 2012 to 31 July We included first 6 months data for 2013 with an allowance of 1 month for late comers (up to July 2013) as anyone with a viral load then had been on treatment for at least 6 months which extends back to end of 2012.
Mid-year Population Estimates 2012, Statistics SA
National HIV and Behavioural Survey, 2012, Human Sciences Research Council

5. HIV diagnoses, by mode of transmission, 2005-2014, EU/EEA
HIV/AIDS Surveillance in Europe 2011
HIV diagnoses, by mode of transmission, , EU/EEA
Data is adjusted for reporting delay. Cases from Estonia and Poland excluded due to incomplete reporting on transmission mode during the period; cases from Italy and Spain excluded due to increasing national coverage over the period.
Source: ECDC/WHO (2015). HIV/AIDS Surveillance in Europe, 2014

6. New HIV diagnoses, by CD4 cell count per mm3 at diagnosis and region of origin of the case, EU/EEA, 2014
< 200 cells/mm3
200 to <350 cells/mm3
350 to <500 cells/mm3
>500 cells/mm3
Source: ECDC/WHO (2015). HIV/AIDS Surveillance in Europe, 2014

7. In September 2015, ECDC hosted an expert meeting on the HIV continuum of care in Europe. One of the goals
of the meeting was to identify ways to improve the analysis of the continuum across the region.
The experts recommended focusing on four core issues that are essential to improving the HIV response: the
estimated number of people living with HIV, testing/diagnosis, treatment and viral suppression. There was a
parallel recommendation to move toward standardised definitions for each of the stages.
At the October 2015 meeting of the advisory group for monitoring the Dublin Declaration, the decision was
made to monitor the HIV continuum of care using the recommended continuum.

8. Continuum of HIV care: United Kingdom, 2014
In the UK, free and accessible HIV treatment and care has resulted in large-scale treatment coverage: in 2014, an estimated 75% of all people living with HIV (PLWH) (diagnosed and undiagnosed) were treated and 70% of all PLWH (72,800/103,700) had an undetectable viral load (less than 200 copies/UL). This figure is close to the ambitious UNAIDS target of 73% of all PLWH being virologically suppressed, as laid out in the goals (90% of people living with HIV being diagnosed, 90% diagnosed on ART and 90% viral suppression for those on ART by 2020).
Presentation title - edit in Header and Footer

9. Quality of care indicators for adults (aged≥15 years) receiving HIV care: United Kingdom, 2010

10. Two problems with incidence surveillance that need to be solved
New diagnosis is not the same as incident cases
Time from infection to diagnosis can be measured by CD4 cell count at diagnosis
Incidence can be modelled from CD4 cell count at diagnosis*
Challenging to set up a system that captures all newly diagnosed cases
Laboratory data should form the basis of a surveillance system
Preferably automated laboratory data with unique identifiers
* Ard van Sighem, Fumiyo Nakagawa, Daniela De Angelis, et al.
Estimating HIV Incidence, Time to Diagnosis, and the Undiagnosed HIV Epidemic Using Routine Surveillance Data.
Epidemiology Sep; 26(5): 653–660.

11. Linking epidemiological data to laboratory data in Denmark
Positiv HIV-test in laboratory automatically generates form to requesting physician with epidemiological questions to fill in and send to national surveillance team
Transmission mode, country of infection, CD4 and VL at diagnosis, symptoms etc.
Using unique identifiers
minimises time spent cleaning data (e.g. duplicates)
Enables linkage with other databases (deaths, TB etc)
Requires utmost confidentiality
If form not recieved within reasonable time physician gets reminder
Dealing with (often non-random) missing information is controversial
Economic incentives could have saved reminder-work (e.g. no payment for a patient whithout a form)

12. Collected and processed data put to use are key to informing where to put in efforts
Use “real time” monitoring of the epidemic and treatment programs – case reporting by authorities/agencies
Use epidemiological data to monitor transmission dynamics
Link data systems to describe the (current) epidemic
Use CD4 cell counts and HIV viral load to model and monitor late diagnoses, entry to care, and quality of care
Inform efforts to diagnose persons promptly and to reduce HIV transmission
Monitor the HIV continuum or cascade of HIV care
Ensure that HIV care is optimal for all

13. Thank you for your attention