For vaccine - GLOBE NETWORK

How to determine the effectiveness
of Hib and pneumococcal conjugate
vaccines in developing countries?
Kim Mulholland
London School of Hygiene and Tropical
Medicine, and Menzies School of Health
Research, Darwin

We know that pneumococcal
conjugate vaccines are effective ‐
• From RCT’s
– Nth California (PCV‐7, Prevnar®, Wyeth)
– Gambia (PCV‐9, Wyeth)
– S Africa (PCV‐9, Wyeth)
• From post‐introduction surveillance
– USA (PCV‐7, Prevnar®, Wyeth)
– UK, Canada, Australia, etc (PCV‐7, Prevnar®,
Wyeth)

But difficult questions have arisen…
• Uncertain disease burden
– Few data on pneumococcal disease burden in Asia
– Philippines 11‐v PCV trial
• No impact on clinical pneumonia
• Serotype replacement
– Wherever PCV’s introduced, carriage and disease
due to non‐vaccine types (NVT) ↑

But difficult questions have arisen…
• Uncertain disease burden
– Few data on pneumococcal disease burden in Asia
– Philippines 11‐v PCV trial
• No impact on clinical pneumonia
• Serotype replacement
– Wherever PCV’s introduced, carriage and disease
due to non‐vaccine types (NVT) ↑

Bohol study of PCV‐11
• 12,194 infants randomized (individually) to
receive PCV‐11 (Sanofi Pasteur, never
licensed) or placebo at 6, 10 and 14 weeks
• Radiological pneumonia
– PCV‐11 – 93
– Placebo – 120
• Clinical pneumonia
– PCV‐11 – 934
– Placebo ‐ 930
Vaccine efficacy 23%, N/S
Vaccine efficacy 0
PIDJ 2009;28:456-62

Lombok study of Hib vaccine (PRP‐T)
• 55,073 infants randomized (by hamlet) to
receive PRP‐T+DTP or DTP at 6, 10 and 14
weeks
• Radiological pneumonia (rate/1000
chld.years)
– PRPT – 9.4
– Control– 8.9
• Clinical pneumonia
– PRP‐T – 380
– Control ‐ 395
Vaccine efficacy

But difficult questions have arisen…
• Uncertain disease burden
– Few data on pneumococcal disease burden in Asia
– Philippines 11‐v PCV trial
• No impact on clinical pneumonia
• Serotype replacement
– Wherever PCV’s introduced, carriage and disease
due to non‐vaccine types (NVT) ↑

Invasive pneumococcal disease in Alaskan
natives – under 2 years
450
400
350
300
250
200
150
all disease
vaccine type
non-vaccine type
100
50
0
1995-2000 2001-2003 2004-2006
JAMA 2007;297:1784‐92

So the key questions are…
• How to establish the effectiveness of Pnc
vaccines in settings where the contribution of
Pnc to pneumonia burden is in doubt?
• How to detect declining effectiveness that
may be due to serotype replacement?

How to establish the effectiveness of
Pnc vaccines in settings where the
impact is in doubt?
• Before and after?
• Case control study?
• RCT?

Before and after…
• Need consistent and complete case
ascertainment
• Outcomes:
– Invasive pneumococcal disease
– Pneumonia
– Mortality
• Problems
– Viral epidemics
– Changing referral patterns

Cumulative incidence per 1000 population per year, hospitalised
radiologically confirmed pneumonia, by study year and age group,
NT Indigenous children, April 1997 – March 2005
70
60
1-11mths
12-23mths
24-59mths
All children
50
Rate per 1000 population
40
30
20
10
0
Apr97 - Mar98 Apr98 - Mar99 Apr99 - Mar00 Apr00 - Mar01 Apr01 - Mar02 Apr02 - Mar03 Apr03 - Mar04 Apr04 - Mar05
Study year
Data courtesy of Dr Kerry‐Ann O’Grady, Menzies School of Health Research, Darwin

ARI surveillance system
Viet Nam
Target population:
Paediatric cases admitted with ARIs: fast and/or difficulty in breathing
Catchment area: 16 communes of Nha Trang city(total 27 communes)
Khanh Hoa District
Clinical‐epidemiological data:
Demographic, clinical and management data
Sample:
Nasopharyngeal swab for Nucleic acid extraction
Others:
Chest Xray and blood sample
CXR: standardized by MiASoft, Ltd. (Faringdon, UK) radiology training modules.

Number of ARI/Pneumonia cases Nha Trang:
Jan 29th, 2007 to Jun 30th,2008: 1126 cases
Jan 2007 Jan 2008

Seasonal Prevalence of Major Pediatric Viral Respiratory Pathogens
60
Positive rate (%)
50
40
30
20
Rhino
RSV
Inf A
10
0
2007
Feb
2007
Mar
2007
April
2007
May
2007
June
2007
July
2007
Aug
2007
Sept
2007
Oct
2007
Nov
2007
Dec
2008
Jan
N=29 53 87 88 75 136 100 82 78 44 45 64 37 112
2008
Feb
2008
Mar
Slides courtesy of Dr Lay Myint, University of Nagasaki

Cumulative incidence per 1000 population per year, hospitalised
radiologically confirmed pneumonia, by study year and age group,
NT Indigenous children, April 1997 – March 2005
70
60
1-11mths
12-23mths
24-59mths
All children
50
Rate per 1000 population
40
30
20
10
0
Apr97 - Mar98 Apr98 - Mar99 Apr99 - Mar00 Apr00 - Mar01 Apr01 - Mar02 Apr02 - Mar03 Apr03 - Mar04 Apr04 - Mar05
Study year
Data courtesy of Dr Kerry‐Ann O’Grady, Menzies School of Health Research, Darwin

Before and after ‐ conclusions
• Under ideal conditions may be able to
produce plausible estimates of the impact of a
vaccine
• For pneumonia and diarrhoea vaccines –
– Need to monitor all external factors, esp.
epidemics
– Very difficult to get convincing results unless the
effect is very big

Case control studies
• Provide estimate of VE, not vaccine impact
• Fundamental problem – vaccinated children have
different risk patterns to unvaccinated
– Especially for risk dependent conditions ‐ pneumonia
• Key questions:
– Are the cases being studied representative?
– Controls
• Community controls?
• Hospital controls?
• How many?
– How do we know the vaccination status of cases and
controls?

Community controls
• Usually match by neighbourhood or similar
– Hopefully risk of disease more or less similar,
apart from vaccination status
– Risk of overmatching (some villages all vaccinated)
• Main problem is access to care/careseeking
– Other community members may not use health
services in the same way
• Major practical on‐ground problems
– Persons inside do not open door
– No‐one home at houses with young children

Hospital controls
• Details of how selected usually not given
• Deal with the issue of access to care
• Chronic or recurrently hospitalized children?
• Malnutrition?
• Sometimes specify particular conditions:
– Pnc meningitis for Hib meningitis studies
– Gastroenteritis for pneumonia

Potential for bias
• Community controls
– Include non‐health care users
• Against vaccine
– Include “convenient” controls
• For vaccine
• Hospital controls
– Include chronic, hospitalized patients
• For vaccine
– Include malnourished children
• Against vaccine

Hib in Asia
• Perceptions among paediatricians:
– Hib meningitis not important in Asia
• Likely role of antibiotics in preventing meningitis
– Hib and Pnc may not contribute as much to
pneumonia as in Africa or the Pacific
• Becoming less important as vaccine price falls

Lau et al. Acta Paediatr 1995;84:173‐6
• 5 year study covering whole of Hong Kong
• service coverage and bacteriology good
• 57 cases detected:
– incidence of Hib meningitis 1.75/10 5 /year
– incidence of invasive Hib disease 2.67/10 5 /year
ie. lifetime risk 0.013%
(150 fold less than Africa)

Bacterial meningitis in Beijing Children’s Hospital, 1955-2000
DEATHS
60
50
40
30
20
10
0
NM outbreak
DEATHS
NM vaccination
ADMISSIONS
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
The past 35 years have seen a considerable decline in patient numbers and case
fatality (Previously unpublished data)
Data courtesy of Prof HK Yang
ADMISSIONS
2000
1800
1600
1400
1200
1000
800
600
400
200
0
year

An Effectiveness Trial of Hib Vaccine in
Bangladesh
• Objective:
– Determine proportion of pneumonia and meningitis in

Potential for bias
• Community controls
– Include non‐health care users
• Against vaccine
– Include “convenient” controls
• For vaccine
• Hospital controls
– Include chronic, hospitalized patients
• For vaccine
– Include malnourished children
• Against vaccine

How to establish the effectiveness of
Pnc vaccines in settings where the
impact is in doubt?
• Before and after?
• Case control study?
• RCT?

An RCT of a vaccine of proven efficacy
to determine the vaccine‐preventable
burden of disease? (“vaccine probe”)
• Ethical issues
• Financial issues
• Study quality

Lancet 2005;365:43‐52

Pneumonia results
Numbers of events in children less than 2 years
Outcome
Control
group
Hib vaccine
group
Clinical pneumonia 6179 6273
Any severe
932 975
pneumonia
Pneumonia
1651 1657
hospital admission
X-ray confirmed 330 361

Pneumonia results
Incidences per 100,000 child-years in children less than 2 years
Outcome
Control
group
Hib vaccine
group
Preventable
disease incidence
Clinical pneumonia 39,516 37,954 1,561 (270 – 2,853)
Any severe
5,460 5,196 264 (-101 - 629)
pneumonia
Pneumonia
4,518 4,345 -174 (-153 - 500)
hospital admission
X-ray confirmed 893 936 -43 (-185 - 98)
Fatal pneumonia 1020 954 66 (-126-259)

Meningitis results
Incidences per 100,000 child-years in children less than 2 years
Outcome
Control
group
Hib vaccine
group
Preventable
disease
incidence
Laboratory
19 2.6 16 (1.4 – 31)
confirmed Hib
Purulent bacterial 86 39 47 (13 – 81)
meningitis
All meningitis or 701 543 158 (42 – 273)
acute seizures*
Meningitis deaths 161 121 40 (-17 – 97)
* Many of these children were not admitted

Consider country X
• East Asia
• Moderate mortality, pneumonia a big problem
• Under pressure to introduce pneumococcal
vaccine
• Wary about
– IPD burden
– Burden of pneumonia due to Pnc
– Serotype issues
– Future cost of the vaccine
• Country needs better evidence

Phased introduction
• In part or all of the country
• Stage 1
– At least 30 administrative units randomized to receive
or not receive vaccine
– Annual analysis of pneumonia (and possibly mortality)
rates
• Compare rates of disease in vaccinated and unvaccinated
sectors
• Support with case control study
• Stage 2
– Introduce vaccine into other areas
– Consider introduction of second vaccine and continue
analysis

Phased introduction
• Risks
– May delay vaccine introduction
– Interim analyses may produce misleading results
– Measured effectiveness may decline later due to
serotype replacement
– Groups may be unbalanced with respect to:
• Risk of disease
• Access to health care

Phased introduction
• Advantages
– Real data for country and region
– Accurate cost‐effectiveness analysis
– What alternatives are there?
• Blind introduction?

What of serotype replacement?
• Options
– Ignore and hope for the best
– Monitor carriage patterns
– Monitor IPD in sentinel sites
– Monitor IPD in high risk groups
– Try to monitor impact on pneumonia rates

Monitoring IPD
• May indicate when replacement has
neutralized effectiveness (eg. Alaska)
• May miss this altogether
– Pnc serotypes vary with respect to:
• Disease patterns
• Probability of disease given carriage
• Probability of bacteraemia given disease
– In US replacement was identified because 19A
was dominant replacing serotype and quite likely
to cause bacteraemia

IPD in high risk groups
• Immunocompromised are more susceptible to
disease due to replacement serotypes
• Monitoring IPD in HIV+ may be a useful way of
monitoring replacement in settings with many
HIV+ cases (eg. S Africa)

• Hib
Conclusions
– Burden issues in Asia less important as vaccine is
becoming inexpensive
• Pneumococcal
– Burden issues important due to vaccine cost and
extra injection
– Monitoring effectiveness essential
• Case control studies and sentinel monitoring
of IPD alone is inadequate
• Monitoring for impact of replacement must
involve monitoring pneumonia rates