Optimal Response to Threatened Reintroduction of Smallpox: Design via Mathematical Modeling

1
Optimal Response to Threatened Reintroduction of
Smallpox Design via Mathematical Modeling

• Background, NIH Consultation,
• Epidemiology John Glasser
• Economics Martin Meltzer

2
Modelers

• Aim to assist in designing, evaluating and
  improving health policy, which has usually been
  made w/o such assistance,
• Human conceptual limitations are such that we
  simplify reality to make it manageable
• Sometimes we over-simplify (i.e., omit or
  mis-specify important factors, processes, )
• Computers reduce the need to simplify, and hence
  this risk, enabling us to deal with complex
  systems approximating reality

3
Modelers (contd)

• Employ myriad approaches, each with strengths and
  weaknesses, all attempting to model disease in
  human populations
• Some are better than others for particular
  purposes, but none is perfect
• Try alternative interventions, all else equal
  (i.e., experiment), compare results to determine
  which attains the objective or
• uses available resources most effectively, and
  which require more or different resources

4
NIH Consultation

• Purpose the idea of this consultation is to get
  all of the major smallpox modelers in the same
  room, talking with each other and with several
  key smallpox biology/epidemiology experts and
  BT-response types. Hopefully, this will lead to
  cross-fertilizations and collaborations - and
  thus to better models.

5
Smallpox Modeling

• Some model populations comprise individuals,
  others groups
• Most models contain unrealistic biological or
  logistical assumptions, but
• Several deduce consequences of simple rule-based
  individual decisions
• One explicitly evaluates how logistical
  considerations influence strategy
• Modelers agree that heterogeneity and chance
  affect disease transmission

6
I promised

• to facilitate communication
• of policy questions to modelers
• ideas and results to policymakers
• I prefer to do this via, or together with, the
  consultation organizer

7
Please see abstract
8
CDCs model features

• Broadly-overlapping normal and modified disease
  spectra for, respectively, immunologically-naïve
  people and those with residual immunity or who
  have been vaccinated, but not soon enough
  post-exposure to avert disease
• Prophylactic or post-event mass vaccination,
  tracing and vaccination of contacts, isolation of
  cases and dissemination-stage contacts
• Artificially-induced and naturally-acquired
  immunity that wane at different rates
• Boosting via exposure or vaccination
• Seasonally-varying infection rates

9
Outline

• Historical re-introductions (fit adjusting only
  infection rates)
• Stockholm, 1963
• Bangladesh, 1972
• Policy questions
• prophylactic vs post-event
• contact tracing and isolation, ring or mass
  vaccination
• Refinements
• spatial-, age- and social-structure
• event-driven stochasticity

10
Stockholm, 1963

• Typical of several re-introductions to Europe
  from southern Asia (e.g., institutional, small
  only 27 cases and 4 deaths, )
• But vaccination status and other clinical
  details permit assignment to disease spectra
• And the response, which extended far beyond the
  affected community, is informative
• 8,000 neighbors, and roughly 300,000 other
  residents were vaccinated, of whom 1,076 reported
  complications, 77 serious

Thanks to Peter Merkle and John Bombardt for
sharing the original articles.
11
(No Transcript)
12
Bangladesh

• No cases of smallpox were reported during 1971,
  possibly indicating elimination
• But 10 millions fled to India during what was
  East Pakistans struggle for independence
• Some were infected in 4 refugee camps and
  returned incubating or infectious
• Magnitude of the problem not immediately apparent
  by virtue of disrupted infrastructure

13
Smallpox in Bangladesh
14
Smallpox Scar Survey, 1976

• Sampled age group within which known proportion
  of cases occurred
• Counted people with pockmarks, corrected for
  mortality due to disease,
• fading of pocks and all cause mortality during
  the intervening period
• Inflated by reciprocals of sampling and age group
  fractions

15
(No Transcript)
16
Modeling Process

• Model is faithful to our collective understanding
  of smallpox transmission in human populations
• Fits historical datasets reasonably well with
  informed opinions about parameters that can no
  longer be estimated independently
• Age-, spatially-, and socially-structured models
  would be informative (i.e., increase
  understanding of biology), but

17
Experiments

• Baseline residual immunity, varied introduction
  magnitude and R0
• Prophylactic varied coverage with and without
  regard for residual immunity
• Post-introduction
• Contact tracing, mass vaccination
• Increasingly sensitive definitions of contacts
  (resulting in misclassification of susceptibles)
  ? ring vaccination
• Combinations

18
Summary

• Introduction magnitude and R0 (a surrogate for
  host density, movement, ) affect mostly outbreak
  timing. Artificially-induced immunity affects
  magnitude as well
• Mass vaccination works well, especially if R0 is
  modest and response is timely and aggressive
• Contact vaccination and isolation can be very
  effective, even if tracing efficiency and
  compliance are modest, as cases also are
  isolated, and their compliance is excellent