The potential impact of different

vaccination policies
 
John  Edmunds,  Anton  Camacho,  Roz  Eggo,  Adam  Kucharski,  
Sebas;an  Funk,  Alexis  Robert,  Conall  Watson,  Stefan  
Flasche  
 
London  School  of  Hygiene  and  Tropical  Medicine  
 
john.edmunds@lshtm.ac.uk  
Ring vaccination
All  individuals,  i.e.  Including  unvaccinated  

•  Randomised  clinical  trial   Delayed  

gives  clearest  evidence  of  
effec;veness  of  ring  
vaccina;on  policy   Early  
•  Are  there  condi;ons  when  
it  might  fail  to  control  an  
outbreak?  

Henao-­‐Restrepo  et  al.  Lancet  (2015)  

Ring vaccination

Transmission
model  

Vaccination delays  
80% efficacy and 70%
vaccinated means:
Rv = (1-0.7*0.8)Rw = 0.44Rw  
 • Initial (i.e. missed) cases have R=7
• Or R=2.5 if individuals with funeral
and hospital transmission omitted
• Other cases have R=0.66

Faye et al (2015) The Lancet Infectious Diseases  

If missed cases associated with superspreading, ring 0= No vaccination
vaccination might fail to stop large outbreak (i.e. >500 1= Ring vaccination
clusters) even if small proportion of cases missed 2= Mass vaccination  

If no funeral or hospital superspreading events, ring vaccination

can prevent large outbreaks even if ~40-50% cases missed  
 Model for mass vaccination

-­‐  Mass  vaccina;on  at  the  regional  level  

-­‐  Liberia:  county  
-­‐  Sierra  Leone:  district  
-­‐  Meta-­‐popula;on  model  
-­‐  Importa;ons  from  neighbouring  regions  
-­‐  Rate  dependent  on  size  &  propor;onal  to  cases  
 Base case results 1500
Liberia
no vaccination
Vaccina;on  (base  case)  

Incidence
1000 vaccination
80%  efficacy  
500
Trigger:  10  cases  per  region  
2  week  delay  to  start   0
0 10 20 30 40 50
100,000  doses  per  region   Weeks

per  week  
800
Sierra Leone
no vaccination
70%  coverage    
600

Incidence
vaccination
400

200

0
0 10 20 30 40 50
Weeks
15000
Liberia   Sierra    
Leone  
total cases

10000

5000

0
cc

cc
in ve

in ve
n

n
va

va
io

io
cc cti

cc cti
at

at
no

no
va rea

va rea
 Delay in vaccine availability 1500
Liberia
no vaccination

Incidence
1000 immediate
from 1 Aug 2014
500 from 1 Oct 2014

0
0 10 20 30 40 50
Weeks
800
Sierra Leone
600 no vaccination
Incidence

immediate
400 from 1 Aug 2014
from 1 Oct 2014
200

0
0 10 20 30 40 50
Weeks
Cases  Averted   Doses  Deployed  
Sierra  Leone   Sierra  Leone  
10000
Liberia   3.5 mil

8000 3 mil

Liberia  
cases averted

doses deployed

2.5 mil
6000
2 mil

4000 1.5 mil

1 mil
2000
0.5 mil

0 0 mil
g

g
c

c
e

e
ct

ct

g
c

c
e

e
ct

ct
c

c
Au

Au

c
Au

Au
t

t
O

O
ia

ia
O

va

va
ia

ia
va

va

ed

ed
1

1
1

1
ed

ed

no

no
1

1
1

m
no

no

im

im
m

m
im

im
HCW: questions
•  Do  HCW  play  an  ac;ve  role  in  driving  the  transmission  during  
the  increasing  phase  of  the  epidemic?  
•  What  would  have  been  the  benefit  (direct  and  indirect)  of  
vaccina;ng  HCW  before  this  epidemic?  

LocaAon   PopulaAon  size   Confirmed  &  probable  cases  

Community   HCW   Community   HCW  
Kikwit  (1995)   200,000   900   241   76  
Liberia  (2014)   4.3  m   11.7  k   5  k   378  
Sierra  Leone  (2014)   6.3  m   6.2  k   9  k   307  
Guinea  (2014)   10.5  m   1.7  k   3.7  k   195  
 Transmission tree (Kikwit)
health-­‐care  worker  
1  HCW   community  
39  secondary  cases  
25  (69%)  are  HCW  

1  COM  
21  secondary  cases  
100%  are  COM  

262/316  (83%)  known  index  cases    

Vaccination
•  Vaccine  type:  single  dose.  
•  Vaccine  efficacy:  70,  80  or  90%.  
•  Protec;ve  1  week  post  vaccina;on.  
•  All-­‐or-­‐nothing  immunity.  
•  Immunity  lasts  for  1  year.  
•  Scenario  1:  vaccinate  all  health-­‐care  workers  before  the  epidemic.  
•  Scenario  2:  vaccine  campaign  in  the  community:  
–  100,000  doses  per  week  (2  weeks  to  vaccinate  all  Kikwit  area)  
–  Starts  on  20  April  (2  weeks  afer  the  ini;al  case  in  Kikwit  
General  Hospital)  or  on  10  May  (arrival  of  interna;onal  
response  team)  
•  Scenario  3:  scenario  1  +  2  
Proportion of cases averted
Comparing  Scenario  1  and  Scenario  3,  Timing  of  campaign  
com hcw
100 100
proportion of cases averted

●
●
● ●
●
● ●
●
●
● ● ●
75 75 ●
●
●
●
●
●
●
●
●
●
●
● ●
● ●
●
● ● ●

50 50
●
●

25 25
●
●
● ●
●
●
●
0 0

20 April 10 May Only HCW 20 April 10 May Only HCW

Vaccine efficacy
community  cases  can  be  averted  
•  Later  starts  to  the  campaign  (>34  days)  result  in  liile  addi;onal  benefit  to  vaccina;ng  
only  HCW  before  the  epidemic  
Summary
General  approach  
–  Impossible  to  tell  what  next  epidemic  will  be  like  
–  What  if  vaccine  had  been  available  in  past  outbreaks  
Ring  vaccinaAon  
–  Trial  demonstrates  the  effec;veness  of  this  strategy  
–  Least  effec;ve  if  cases  who  “escape  detec;on”  have  high  reproduc;on  number  
–  May  need  to  widen  ring  &/or  supplement  with  more  widespread  vaccina;on  
•  Stockpile  implica;ons  
Mass  vaccinaAon  (district,  country,  etc)  
–  Effec;veness  of  vaccina;on  depends  on  ;ming  
•  Late  vaccina;on  has  liile  impact  
HCW  vaccinaAon  (prophylacAc)  
–  HCW  at  very  high  risk,  par;cularly  at  the  outset  of  Ebola  epidemics  
–  May  also  play  a  role  in  amplifying  ini;al  spread  
•  Vaccina;on  of  HCW  has  poten;al  popula;on-­‐level  effects  
General   approach  
–  i.e.  limi;ng   spread  to  the  community  
Acknowledgements
& further details
More  detailed  weekly  assessments  and  district-­‐level  forecasts  at:  
hip://cmmid.lshtm.ac.uk/research/ebola/  
 
Data:  
•  MoHs  
•  WHO  
•  MSF  
 
Funded  by: