Developments in infectious disease surveillance - Dr. Jeff Zimmerman, Vet Diagnostic and Production Animal Medicine, Veterinary Diagnostic Laboratory, Iowa State University, from the 2016 North American PRRS Symposium, December 3‐4, 2016, Chicago, Illinois, USA.
More presentations at http://www.swinecast.com/2016-north-american-prrs-symposium
Developments in infectious disease surveillance through representative sampling
1. Developments in infectious disease surveillance
M Rotolo, Y Sun, C Wang, LG Giménez-Lirola, R Main, J Zimmerman
2. Definitions
• Surveillance = detection.
• Monitor = track changes
over time.
• In practice, “surveillance”
and “monitoring” are used
interchangeably.
Rodger Paskin. 1999.
3. Surveillance: Humans
• "Representative sampling" described in 1895.
(Kruskal and Mosteller, 1980).
• Normal practice was to sample everybody.
• Lesson from 1948 U.S.
presidential election ...
use statistically-based
sampling. use
7. Surveillance: PRV Official "Random" Samples
• 95/10 (95% probability of
detecting 10% infection)
< 100 pigs - test 25
100-200 - test 27
201-999 - test 28
≥1,000 - test 29
• 95/5 (95% probability of
detecting 5% infection)
< 100 pigs - test 45
100-200 - test 51
201-999 - test 57
≥1,000 - test 59
• 95/20 sampling: up to 14 head, test all. Over 14 head, test 14
8. Surveillance: PRV Official "Random" Samples
• 95/10 (95% probability of
detecting 10% infection)
< 100 pigs - test 25
100-200 - test 27
201-999 - test 28
≥1,000 - test 29
• 95/5 (95% probability of
detecting 5% infection)
< 100 pigs - test 45
100-200 - test 51
201-999 - test 57
≥1,000 - test 59
• 95/20 sampling: up to 14 head, test all. Over 14 head, test 14
11. Based on specific assumptions ...
1. Finite population.
2. Binary outcome (yes/no).
3. Each observation is independent.
4. Target is randomly distributed
in the population.
12. If the assumptions hold, sample size (n) can be
calculated as ...
𝑛 = (𝑁𝑧^2 𝑝𝑞) / ((𝐸^2 (𝑁−1)+𝑧^2 𝑝𝑞))
• n = Minimum sample size for detection
• N = Population size
• z = Confidence level (zα/2)
• p = Proportion of events in population
• q = Proportion of non-events in population
• E = Accuracy of sample proportions
13. Need for evolution of surveillance ...
• PROBLEM: Complexity of systems
• 2012: 60% of inventory on farms
with ≥ 5,000 head.
• RESPONSE: New sample types
• Individual pig samples
• Pooled samples
• Aggregate samples (OF, air, water,
Swiffers)
16. Recipe for collecting oral fluids:
1.Collect first thing in the
morning (pigs are most active)
2.Use cotton rope
3.Adjust rope to pig size
4.Extract fluid from rope
5.Pour fluid into a tube chill
or freeze
6.Send for testing
18. PCR detection of pathogens in oral fluids
• African swine fever virus
• Classical swine fever virus
• Foot-and-mouth disease
virus
• Influenza viruses
• PCV2
• Aujeszky’s disease virus
• PEDV
• PRRS virus
• TTV1 and 2
• Erysipelothrix spp.
• M. hyorhinis
• M. hyosynoviae
19. Possible oral fluid antibody assays ...
• African swine fever virus
• Aujeszky's disease virus
• Classical swine fever virus
• FMD virus
• Influenza viruses
• PCV2
• PEDV
• PRRSV
• Actinobacillus
pleuropneumoniae
• Erysipelothrix spp.
• Salmonella spp.
• Lawsonia intracellularis
• Any pathogen for which
we have a good serum
ELISA
20. How to collect oral fluid samples in a
“statistically valid” surveillance approach?
21. Statistically valid method to collect OF samples?
• Allocation of samples
• Number of samples
• Frequency of sampling
BARN LEVEL -
NOT SITE LEVEL
22. Statistically valid method to collect OF samples?
• Allocation of samples?
• Number of samples?
"piecewise exponential survival model"
23. Field data
• 3 wean-to-finish barns on one site
• 36 pens per barn (~25 pigs per pen)
• Sampling from the end of the first
week placed + weekly for 8 weeks
(9 samplings in total)
24. Field data
• ~972 oral fluid samples
• 3 barns x 36 pens x 9 samplings = 972
• All samples randomized and then tested
• PRRSV RT-PCRs were used in the analysis
30. How to collect oral fluid samples in a
“statistically valid” approach?
31. Statistically valid method to collect OF samples?
• Allocation of samples?
• Number of samples?
"piecewise exponential survival model"
32. "Allocation" of samples?
"Fixed" spatial sampling
Equidistant spacing in
buildings and/or spaces.
Random sampling
Simple random sampling
(random.org).
36. Statistical analysis: fixed spatial sampling
was EQUAL OR BETTER than random samling
Detectionbyprevalence
(100%dxse,100%dxsp
37. • Easily understood and applied
• Widely used, e.g., forestry and
environmental sciences.
• "Systematic sampling is more precise than simple
random sampling when spatial autocorrelation is
present and the sampling effort is equal." (Aune-
Lundberg and Strand, 2014)
Systematic spatial sampling
38. Poster 23. Rotolo et al. Spatial
autocorrelation and implications for oral fluid
based PRRSV surveillance.
&
Tuesday 8:00 am CRWAD
39. Detection on a site depends on the number of
barns sampled!
40. Probability of detection (P) as a function of the
number of barns on the site
P = (1 - (1 - p)n) n = number of barns sampled
Probability (p) of detection (one barn)
XXXXXXXXXXXXX
41. Probability (p) of detection (one barn)
Probability of
detection (site)
XXXXXXXXXXXXX
43. Fixed spatial sampling works for monitoring …
10 sites x 6 pens in each barn x sampling each
2 weeks for 18 weeks.
44. 0
0.5
1
1.5
2
2.5
3
3.5
0 2 4 6 8 10 12 14 16 18
MeanS/Pratio
Weekof Growout
Barn 10
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
0 2 4 6 8 10 12 14 16 18
MeanS/Pratio
Weekof Growout
Barn 5
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
0 2 4 6 8 10 12 14 16 18
MeanS/Pratio
Week of Growout
Barn 9
P
-0.50
0.00
0.50
1.00
1.50
2.00
2.50
3.00
0 2 4 6 8 10 12 14 16 18
MeanS/Pratio
Week of Growout
Barn 6
P
P
P
P
P
P
0
0.5
1
1.5
2
2.5
3
3.5
0 2 4 6 8 10 12 14 16 18
MeanS/Pratio
Weekof Growout
Barn 10
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
0 2 4 6 8 10 12 14 16 18
MeanS/Pratio
Weekof Growout
Barn 5
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
0 2 4 6 8 10 12 14 16 18
MeanS/Pratio
Week of Growout
Barn 9
P
-0.50
0.00
0.50
1.00
1.50
2.00
2.50
3.00
0 2 4 6 8 10 12 14 16 18
MeanS/Pratio
Week of Growout
Barn 6
P
P
P
P
P
P
Results (averages)
ELISA S/P values
RT-PCR positives (P)
45. The results are logical and easy to understand because they
reflect the pigs' response to infection over time
Results (averages)
ELISA S/P values
RT-PCR positives (P)
46. This exmple used PRRSV PCRs, but the model
applies to any pathogen / assay combination.
47. 1. Sampling recommendations should
apply to any oral fluid-based test.
2. Use fixed spatial allocation
3. Works for either detection
(surveillance) or monitoring
4. Approach compatible with
regional disease control projects.
CONCLUSIONS - Oral fluid surveillance
48. Improving our response to
swine health challenges
M Rotolo, Y Sun, C Wang, LG Giménez-
Lirola, R Main, J Zimmerman