1. Serosurvey among TB patients Sampling. Design. Sample size
Serosurvey among TB patients Sampling Design Sample size Timeframe Operational procedures Staffing and training Data collection and management Specimen collection and testing Data collection and management Quality assurance Measures to protect anonymity
SENTINEL SURVEILLANCE SYSTEMS
Dr. GEORGE LOTH, SIR/HIV. NOV 2004

2. Review of Design Options for HIV Surveillance
Passive surveillance of reports of HIV positive test results from laboratories, institutions, and hospitals where HIV testing is done
Active surveillance by population-based sample surveys or cross-sectional serosurvey
Studies such as seroincident cohort studies to monitor changes in HIV infection in identified individuals who return periodically for testing
Active surveillance of specific populations at sentinel sites using unlinked anonymous testing procedures
THIS IS RECOMMENDED AS THE BEST OPTION #

3. Steps for Setting up an HIV Sentinel Surveillance System: General Survey Methods
Selection of sentinel populations
Selection of sample inclusion criteria
Selection of sites for sentinel surveillance
Review of methods for collecting blood samples
Review procedures for maintaining confidentiality
Determine data to be collected with blood samples
Determine methods for compilation, analysis, presentation, and dissemination of collected data at central and local levels
# 3-A-3

4. Considerations when selecting TB clinics
The selected TB clinics should not only be able to obtain the sample size needed for the serosurvey, but they should also be able to provide a national picture of the epidemic, which entails obtaining information from:
• different geographical locations, including those with a high risk of HIV infection (e.g., borders, transport corridors);
• areas with different population densities and sizes;
• both urban clinics and rural clinics;
• patients with different socioeconomic status.
SENTINEL SURVEILLANCE SYSTEMS
Dr. GEORGE LOTH, SIR/HIV. NOV 2004

5. Sampling terms
Population: Group that is being sampled and targeted: TB patients
Sample design: How a sample is selected from the population (e.g., convenience sample, simple random sample, cluster sampling with probability proportional to size)
Sample unit: Unit for listing and selecting in a serosurvey (e.g., TB clinic, TB patients)
Sample frame: List of all sample units from which a sample is drawn (e.g., list of all TB clinics, villages, or households in a province or country)
SENTINEL SURVEILLANCE SYSTEMS
Dr. GEORGE LOTH, SIR/HIV. NOV 2004

6. Criteria for selecting TB clinics
• A reliable laboratory for processing of specimens and transport to the laboratory that will be conducting HIV testing.
• The site is accessible to surveillance staff.
• On-site staff members are willing to cooperate and are trained to conduct the serosurveys.
• The site provides services to a sufficiently large number of clients.
• On-site HIV counselling and testing services or referral to such services should be available to clients included in the serosurvey.
SENTINEL SURVEILLANCE SYSTEMS
Dr. GEORGE LOTH, SIR/HIV. NOV 2004

7. Steps for Setting up an HIV Sentinel Surveillance System: Laboratory Testing
Review recommended UNAIDS/WHO HIV testing strategy
Select HIV tests for surveillance
Develop HIV testing protocol for local and central level
Develop quality assurance plan for laboratory HIV testing at the central and local level
# 3-A-5

8. Steps for Setting up an HIV Sentinel Surveillance System
Budget for personnel and equipment needs
Plan for dissemination and presentation of results
National AIDS Committee
Ministry of Health/Government
Media and general public
Sentinel sites/districts/regions
High risk groups
Plan of action and timeline for implementation
# 3-A-10

9. Staffing Local Level (clinic) Regional Level National Level
Clinic staff (nurse or laboratory technician)
Laboratory technician
Supervisory staff (nursing supervisor or senior laboratory technician)
Courier (someone from MOH or laboratory technician)
Regional Level
Survey Coordinator
National Level
Data manager/statistician
Surveillance coordinator
# 3-6-3

10. Steps for Setting up an HIV Sentinel Surveillance System: Training Requirements
Surveillance personnel
Sentinel site staff
Laboratory staff
Supervisory personnel
Data management and analysis personnel
# 3-A-6

11. Steps for Setting up an HIV Sentinel Surveillance System: Supervision of Surveillance Staff
Develop quality assurance plan for sentinel surveillance system
Sentinel sites
Laboratory quality assurance
Logistical issues
Data compilation, analysis, and presentation
Interpretation, use and feedback of results
District, regional, and national staff
# 3-A-7

12. Sampling in HIV Sentinel Surveillance
Definition of sampling
Sample size
Sampling methods
Duration
Frequency
# 3-4-3

13. Definition of Sampling
Process of selecting and studying a
representative subset of a larger
population to estimate an unknown
characteristic or parameter (in this
case HIV prevalence) in that larger
population
# 3-4-4

14. Sample Size Considerations
In general: Bigger is better!
Limited by:
time needed to reach sample size
financial resources
technical resources (workload for site staff, testing capacity, QA/QC…)
# 3-4-5

15. SENTINEL SURVEILLANCE SYSTEMS Dr. GEORGE LOTH, SIR/HIV. NOV 2004

16. Duration of Sampling Period
Trade off between volume at clinic and desired sample size
The shorter the better (preferably no more than 8 weeks, but this can be extended to 12 if necessary) #

17. Frequency of Sampling
The financial, technical and human resources must be available at sentinel site
Conduct once a year
Areas of low prevalence (<2%) or generalised epidemics: every 1-2 years
Use:
Same sites
Same methods
Same time of year #

18. Data Collection
Assure collection of information needed for surveillance
Assure that HIV test results cannot be linked to person
Use appropriately designed data forms
Data forms and lab specimens must be labeled with unique code unrelated to other identifiers
# 3-7-2

19. Flow of Data Collection Forms: Approach A
Sentinel Site
National
Surveillance
Program’s
Data Manager
Add demographic data to top portion of form; remove and send to Data Manager. Send lower portion to Laboratory.
Regional/National
Laboratory
Add HIV test result to lower portion; send form to Data Manager
Add
demographic
data
Demographic
data
Add HIV
test result
HIV
test result
# 3-7-4

20. Analysis and Interpretation of Data
Summarize results (frequency distribution, proportions, prevalence) for:
Entire survey sample
Each site
Each sub-group
Presentation of HIV prevalence by age groups especially helpful
HIV infection in younger persons may represent those with more recent infection
Trends in age groups over time
# 3-7-7

21. Prevalence of HIV in culture-confirmed TB patients (n = 4 639)
Province
HIV prevalence
Free State
71.9%
Mpumalanga
67.2%
North West
66.0%
Kwazulu-Natal
64.4%
Gauteng
63.8%
Limpopo
52.4%
Eastern Cape
30.5%
Western Cape
28.2%
South Africa*
55.3%
* SA rate weighted by Province

22. Dots: observed; line: data fitted to a double logistic curve.
HIV prevalence (%)
Prevalence of HIV in tuberculosis patients in the TB Centre in Chiangmai.
Dots: observed; line: data fitted to a double logistic curve.

23. Selection of Sentinel Sites: Conclusions
Goal is to eventually extend the number of sites to a broad distribution of geographic areas
However, number should not be unmanageable (resources should not be overextended)
It is better to have well conducted surveys in a few locations than to have poorly conducted surveys in many locations
# 3-3-7

24. Site Selection Criteria Requirements
Sites provide services for selected sentinel populations
Blood is drawn from patients/clients as part of routine care or service provided at the site
# 3-3-3

25. Important Characteristics of Sentinel Populations
Definable and easily identified
Obtains health care or services at facilities that draw blood as part of the routine medical services
Minimises participation bias with unlinked anonymous testing
Readily accessible to surveillance staff
# 3-2-3

26. General Recommendation
Sampling Methods:
General Recommendation
Of all three methods, consecutive
sampling is recommended because
selection bias is reduced and it is
simple to execute #

27. Unlinked Anonymous Testing (with informed consent)
Coded specimen
All personal identifiers removed
No counseling required
Used when specimens are collected solely for surveillance purposes
# 3-5-8

28. Specific HIV Testing Methods
Unlinked anonymous (with or without informed consent)
Linked confidential
Linked anonymous
Mandatory testing
Compulsory testing
# 3-5-3

29. Anonymous vs. Confidential Testing
Anonymous: No one knows who the patient is (staff can not identify individual with test results)
Confidential: Staff have access to patient identifying information but do not release information to anyone but patient
# 3-5-4

30. Linked vs. Unlinked Testing
With linked testing, specimen can be traced back to a patient’s chart (or personal identifying information)
With unlinked testing, bond between chart (or personal identifying information) and specimen is broken
# 3-5-5

31. Unlinked Anonymous Testing (without informed consent)
Coded specimen
All personal identifiers removed
No counseling required
Conducted in clinics where left-over blood (originally taken for other reasons, such as syphilis testing) can be tested for HIV surveillance purposes
# 3-5-7

32. Linked Confidential Testing (with informed consent)
Coded specimen
Code linked to personal identifier
Staff know specimen’s identity
Consent and counseling required
Useful when specimens are collected solely for HIV surveillance, for example, among populations not accessible through clinics (sex workers, drug users, men who have sex with men)
# 3-5-9

33. Linked Anonymous Testing (with informed consent)
No personal identifiers obtained
Coded specimen
Code given to patient
Consent and counseling required
Useful when specimens are collected solely for HIV surveillance among populations not accessible through clinics (sex workers, drug users, men who have sex with men) #

34. Unlinked Anonymous Testing is Recommended for HIV Surveillance
Participation bias is reduced and therefore
accuracy is improved
Privacy is maintained
Logistically feasible and practical #

35. • Age distribution of clients
The following information on site-specific characteristics should be gathered for each of the selected sites:
Number of patients per month as well as proportion who are repeat visitors
• Age distribution of clients
• Geographic locality of clinic (urban, semi-urban, rural)
• Residence of clinic population (urban, semi-urban, rural)
• Description of services provided at clinic (family planning, voluntary counselling and testing, prevention of mother-to-child transmission).
SENTINEL SURVEILLANCE SYSTEMS
Dr. GEORGE LOTH, SIR/HIV. NOV 2004

36. Steps for Setting up an HIV Sentinel Surveillance System: Equipment Needs
Specimen collection, serum separation, storage, and transport
Laboratory HIV testing
Supervision
Data compilation, analysis, and presentation
General office equipment and space needs
# 3-A-9

37. Site Selection Criteria Accessibility
A reliable laboratory available on-site or nearby to perform the routine laboratory tests as well as processing the residual blood for later serologic HIV testing
Sites should be readily accessible to national-, regional- and district-level surveillance staff
# 3-3-4

38. Flow Chart of Unlinked Anonymous Testing#

40. SENTINEL SURVEILLANCE SYSTEMS Dr. GEORGE LOTH, SIR/HIV. NOV 2004

41. Appropriate responsibilities for seroserveys personnel
SENTINEL SURVEILLANCE SYSTEMS
Dr. GEORGE LOTH, SIR/HIV. NOV 2004

42. Appropriate responsibilities for seroserveys personnel
SENTINEL SURVEILLANCE SYSTEMS
Dr. GEORGE LOTH, SIR/HIV. NOV 2004

43. Appropriate responsibilities for seroservey personnel
SENTINEL SURVEILLANCE SYSTEMS
Dr. GEORGE LOTH, SIR/HIV. NOV 2004

44. Maintaining motivation among serosurvey personnel
• Develop a sense of serosurvey ‘ownership’
• Clearly define responsibilities and roles for all staff involved, at all levels
• Emphasize the importance of each person’s contribution to the serosurvey’s success
• Provide adequate staff training
• Ensure that the necessary equipment is available to conduct the serosurveys
• Assign certain data management and analysis responsibilities to regional coordinators.
• Provide feedback on staff performance and serosurvey results
SENTINEL SURVEILLANCE SYSTEMS
Dr. GEORGE LOTH, SIR/HIV. NOV 2004

45. HIV status according to gender, previous TB treatment and drug resistance*
HIV positivity
p-value
Gender
Male
Female
51.5%
62.2%
p <
Previous TB Rx
Yes No
56.7%
51.6%
p = 0.134
Drug Resistance
Susceptible
Resistant, not MDR
Resistant, MDR
55.1%
55.8%
60.0%
p = 0.575
* Results weighted by Province

46. Sample Size Determinants
Margin of error = interval width around expected prevalence, e.g. +/-3% (i.e., for expected 10%, accept 7-13%)
Confidence limit (e.g., 90%, 95%, 99%): how confident you are that the true number lies within the interval you’ve measured
To calculate sample size, one can use the following formula*:
N = 4 z² P (1- P) W²
*Where z is the standard normal deviation for a two-tailed , where (1 - ) is the confidence level (for example, since  = .05 for a 95% confidence level, z = 1.96), P is the expected proportion of patients with the outcome, and W is the width of the interval.
# 3-4-6

47. Consecutive Sampling
Sample every eligible person (example: all women aged at first visit) until sample size met or sampling period is complete
Less opportunity for error or unintentional manipulation #

48. Systematic Sampling
Randomly select initial patient who meets eligibility criteria and then select every “Nth” (e.g. 3rd or 5th) eligible patient after that until the sample size is reached
More difficult to execute correctly - requires more attention to procedural detail #

49. Random Sampling
Patients are selected at random (determined by random table or computer program)
Difficult to execute among multiple sites #

50. How to Perform Unlinked Anonymous Testing
Specimen linked when originally taken
Performed in clinic setting where blood is collected regularly for other purposes (usually syphilis testing)
Sera are tested for HIV antibodies after the originally intended lab tests have been run
Personal identifiers removed
Coded with new, unlinked ID number
Collect variables such as age, sex, marital status, geographic area of residence, duration of stay in residence, and occupation #

51. Labelling and Logging
Specimen must be labeled with a code at time of collection and processing
Labels should be placed on the side of tube, not cap
Preprinted cryolables should be used when specimens are stored in cryovials
Surveillance coordinators should provide staff with labels or markers and list of codes
Laboratory logbook or line listing should contain tests results with corresponding code
Logbook should be kept secure; only available to surveillance staff #

52. WHO/UNAIDS Testing Strategies Strategy I
Single test
Surveillance: where prevalence >10%, no results provided
Blood screening
Diagnosis: where prevalence >30% among symptomatic persons #

53. WHO/UNAIDS Testing Strategies Strategy II
Two tests (reactive results confirmed by second assay)
Surveillance: where prevalence  10%
Diagnosis: where prevalence
 30% among symptomatic persons, or
>10% among asymptomatic persons #

54. Selection of HIV Test Must choose testing format:
enzyme linked immunoassay (EIA), rapid EIA testing, or a combination of both
Review operational characteristics of testing technology in relation to country conditions:
laboratory infrastructure, availability of skilled laboratory personnel, and quality assurance and control methods #

55. Steps for Setting up an HIV Sentinel Surveillance System: Personnel Responsibilities
Blood collection, demographic data collection, serum separation, storage
Transport of specimens to laboratory
Laboratory HIV testing at local and central level
Data compilation, analysis, presentation and interpretation
District, regional, and national supervision
# 3-A-8

56. Flow of Data Collection Forms: Two Choices
Demographic and lab data recorded on separate forms
Entered separately into database
Matched by survey number and merged into a single record
Demographic and lab data recorded on same form
Single form sent to data entry where identifiers are removed
# 3-7-3

57. Flow of Data Collection Forms: Approach B
Regional/National
Laboratory
Sentinel Site
National
Surveillance
Program’s
Data Manager
Add demographic data to form; send form to Laboratory
Add HIV test result; send form to Data Manager
Add
demographic
data
Demographic
data
Demographic
data
Add HIV
test result
HIV
test result
# 3-7-5

58. Data Entry Use data entry program with built-in error checks
Use trained data entry staff
Double enter when possible
Enter continuously to avoid large backlog of forms
Check data prior to analysis
Supervisor to review and provide immediate feedback on data collection, entry and management
# 3-7-6

59. Utilising HIV Sentinel Surveillance Data: Programme Planning
Targeting activities for population groups and areas
Identify high-risk groups for targeted prevention programmes
Programme monitoring and evaluation
Long-term impact indicator of programme success
Difficult to use as direct measure of success; use with other data
# 3-8-2

60. Utilising HIV Sentinel Surveillance Data: Advocacy
Resource allocation and short-term planning
Use data to model HIV/AIDS prevalence in population
Use data as baseline information for creating programme goals and objectives
Mobilization of political commitment
Obtain, reinforce or increase commitment of political leaders, health sector, other key sectors, NGOs
Educational value for individual
Mass media or individual counseling may increase awareness
# 3-8-3

61. Considerations affecting the sample size
• Initial or baseline HIV prevalence in a geographic area.
• Magnitude of change in HIV prevalence one wants to be able to detect.
• Accuracy needed for the seroprevalence estimate (width of the confidence interval).
• Degree of certainty that the change in HIV prevalence is statistically significant (level of significance or alpha), usually 0.05.
• Degree of certainty that a statistically significant change, if it occurred, can be detected (power or beta), usually 0.80.
• Percentage of TB patients eligible to be included in the serosurvey. This will involve being able to estimate or determine the patient volume in the TB clinics during the serosurvey period.
• Resources available for conducting the serosurvey (e.g., costs, logistics).
SENTINEL SURVEILLANCE SYSTEMS
Dr. GEORGE LOTH, SIR/HIV. NOV 2004

62. Determine frequency of sampling
Steps for Setting up an HIV Sentinel Surveillance System: Sampling Methods
Determine frequency of sampling
Determine duration of sampling per sampling period
Determine minimum sample size required per sentinel site
# 3-A-4