This document summarizes a capstone dissertation analyzing the impact of response time on infectious disease outbreak outcomes in developing countries. The dissertation used data from 171 outbreaks in the WHO's Global Alert Response database. Key findings included that the average response time was 60.4 days, with slower responses (89 days) for the slowest 25% of outbreaks. Response time did not improve over time. Wealthier countries and those with stronger health systems had faster response times. Yellow fever outbreaks saw much longer response times than other diseases like cholera and meningitis. Response time influenced outbreak size, but other factors were also important. Data gaps and limitations hindered conclusions, calling for an improved global outbreak database.
Impact of Response Time on Infectious Disease Outbreaks
1. Impact of response
time on outcomes of
infectious disease
outbreaks in
developing countries
Capstone Dissertation, Master in Public Health
Stefano Malvolti, Johns Hopkins Bloomberg School of Public Health
2. The recent Ebola crisis stresses the
importance of timely response to outbreaks
Source: http://www.cdc.gov/mmwr/preview/mmwrhtml/mm6435a6.htm
3. This work tests some hypotheses on
outbreak response timing
Outcome of infectious disease outbreaks (measured as
number of cases) is influenced by timeliness of response
Different diseases triggers different response times thus making the impact
of different response times more or less relevant.
High reproduction numbers and long incubation period may make the
timeliness of detection and response more critical (a generalisation of the
prior hypothesis)
Stronger health systems and wealthier countries can afford a slight delay in
response since that answer to the health threat will be more effective and
thus likely to lead quickly to control of the outbreak under
4. Study sample built out of WHO Global
Alert Response database
GAR Database
2295 entries
Single Outbreaks
658
Relevant Single
Outbreaks
420
Outbreaks
Included
171
Combination of
multiple entries
In Endemic Areas
120
In Developed
Countries 67
Single Entry 67
Ongoing 16
Special Cases 1
Non infectious
diseases 5
Not Sufficient Info
229
Single Case 20
identificationScreening&EligibilityInclusion
5. Basic descriptive and univariate statistics
applied to elicit valuable insights
Mean
Median
Correlation Coefficient between delay and cases
Chi-squared Test on transformed categorical distribution
Stratification (cut off points): by disease, reproduction number, Health
Systems strength (physicians/10k inhabitants), wealth (GNI / capita),
incubation period
Transformation: from continuous into categorical with two values based on
cut off points (by disease, R0 = 3, HSS = 0.3 physicians/10k inhabitants,
GNI/capita =1800$, incubation = 7 days)
6. Type and quantity of data trigger
biases and limitation
Absence of many outbreaks into GAR database – selection biases
Secondary data sources - no categorise sources in term of quality and reliability
(15% peer reviewed articles, 15% of data from US CDC or WHO) information
biases
Exclusion based on Incomplete information related to several outbreaks -
selection biases
Use of different sources for different data point for the same outbreak –
analytical biases
Lack of transparency concerning the adherence of the various data sources to
similar definitions for most of the parameters (e.g. index case or date of
notification) - misclassification biases.
Index
Case
Notification
Date
Response
Date
Last Case
Number of
Cases
Number of
Deaths
8. Average response time in developing
world is a major cause of concern
Average response time is 60.4 days
(2 months)
The fastest 25% responses took
approximately 3 weeks (22 days)
The slowest 25% responses took 3
months (89 days)
Up to a maximum of almost 9
months (259 days)
9. Response time has not improved over
time, another reason for concern
OVERVIEW
All Cholera Mening YF
# 171 37 34 30
LD - LC 51% 59% 50% 40%
LD - HC 11% 16% 12% 3%
All LD 63% 76% 62% 43%
HD - LC 30% 11% 26% 33%
HD - HC 8% 14% 12% 23%
DELAY
MEAN 60.4 38.1 63.0 75.9
STD DEV 49.8 46.2 33.9 44.2
MIN 3 3 15 9
1st Quartile 22 12 34.75 42.5
MEDIAN 49.0 16.0 57.0 69.5
3rd Quartile 89 48 91.5 110.25
MAX 259 193 142 165
2
5
10. Better performance of wealthier and strong
health systems countries confirmed
OVERVIEW
#
LD - LC
LD - HC
All LD
HD - LC
HD - HC
DELAY
MEAN
STD DEV
MIN
1st Quartile
MEDIAN
3rd Quartile
MAX
below 0.3 above 0.3 under 1800 above 1800
Weak HS Strong HS Low GNI High GNI
135 36 130 41
45% 75% 48% 61%
13% 3% 14% 2%
59% 78% 62% 63%
33% 17% 29% 32%
8% 6% 9% 5%
64.1 46.5 61.5 44.0
49.6 48.7 61.5 83.0
3 3 3 13.00
25 12 24 16
52.0 26.5 49.0 28.0
92.5 57.25 90 39
259 175 259 141
1
8
7
11. Substantial differences in response time
exist between different diseases
Cholera Meningitis Yellow Fever
12. Delay in response influences number of
cases but other variables play a role
13. Yellow Fever outbreak response
requires much longer time
Substantial difference in response time between diseases
Much longer delays in Yellow Fever response compared to the other
diseases.
Progressive reduction of focus as result of availability of vaccines for
diseases with higher mortality (e.g. Pneumococcal and Rotavirus vaccines)
may have played a role
More limited spread of disease and much smaller number of average cases
per outbreak may be perceived as less threatening by health authorities
and political decision makers.
14. Relevant gaps in availability and
quality of outbreak response data
Absence of a quality and systematic global cross-disease source of data
for outbreak
WHO’s Global Alert Response database where country-reported outbreaks
are meant to be recorded and updated includes only a limited number of
outbreaks and for the one included
key data are often missing,
final updates on the outcome of the outbreaks are almost never recorded,
output from other works (e.g. published papers or reports from other
implementing agencies) are not captured
overall quality of the data can be greatly improved (more updated or different
data can be found not infrequently in other validated sources).
15. More work to do!
Limited number of data point, quality limitations and
limited significance of the analysis hinder ability to draw
conclusions that by clearly identifying drivers of the
problems provides compelling argument for change
extend and further validate the analysis than discuss
emerging insights
Absence of a solid complete and reliable source of
information greatly penalise future efforts aimed at
improving the understanding of the outbreaks and the
best way of addressing them consider the creation of
a global outbreak database