Heath PT, Booy R, Azzopardi HJ, Slack MPE, Bowen-Morris J, Griffiths H, Ramsay ME, Deeks JJ, Moxon ER. Antibody Concentration and Clinical Protection After Hib Conjugate Vaccination in the United Kingdom. JAMA. 2000;284(18):2334-2340. doi:10.1001/jama.284.18.2334
Author Affiliations: Oxford Vaccine Group (Drs Heath and Griffiths, Mrs Bowen-Morris, and Professor Moxon) and the Public Health Laboratory Service Haemophilus Reference Laboratory (Ms Azzopardi and Dr Slack), John Radcliffe Hospital, Oxford, England; Department of Child Health, St Bartholomew's and the Royal London School of Medicine and Dentistry, London, England (Professor Booy); Public Health Laboratory Service Communicable Disease Surveillance Centre, Colindale, London (Dr Ramsay); and the Centre for Statistics in Medicine, Institute of Health Sciences, Oxford (Mr Deeks). Dr Heath is now with the Department of Child Health and St George's Vaccine Institute, St George's Hospital Medical School, London, England.
Context The schedule for Haemophilus influenzae type
b (Hib) vaccination of infants in the United Kingdom consists of 3 doses given
at 2, 3, and 4 months of age. Many countries include a fourth dose (booster)
of Hib vaccine in the second year of life on the basis of declining Hib antibody
concentrations after the primary series. Few data are available to show that
this fourth dose is actually necessary.
Objective To evaluate long-term clinical protection against Hib disease and Hib
antibody concentrations following primary Hib vaccination without a booster
Design, Setting, and Subjects Clinical protection study conducted between October 1992 and March 1999
in the United Kingdom, in which children developing invasive Hib disease despite
vaccination in infancy with 3 doses of Hib conjugate vaccine were reported
by pediatricians through an active, prospective, national survey. Separate
antibody studies were conducted among 2 cohorts of children (n = 153 and n
= 107) vaccinated at 2, 3, and 4 months of age with Hib conjugate vaccine
and followed up to 43 and 72 months of age.
Main Outcome Measures Age-specific vaccine effectiveness, derived from the observed number
of true vaccine failures after 3 Hib vaccine doses compared with the number
of cases expected based on the age-specific rates of invasive Hib disease
obtained prior to the introduction of Hib vaccines; and proportion of children
in the 2 cohorts with Hib antibody concentrations of less than 0.15 and less
than 1.0 µg/mL.
Results Ninety-six true vaccine failures occurring after 3 vaccine doses were
detected. During the study period, an estimated 4,368,200 infants in the United
Kingdom received 3 doses of vaccine; therefore, the vaccine failure rate was
2.2 per 100,000 vaccinees (95% confidence interval, 1.8-2.7 per 100,000).
Although vaccine effectiveness declined significantly after the first year
of life (P<.001), it remained high until the sixth
year of life (99.4% in children aged 5-11 months vs 97.3% in those aged 12-71
months). The proportion of cohorts 1 and 2 with anti-PRP antibody levels of
less than 0.15 µg/mL increased between 12 and 72 months of age (6% at
12 months, 8% at 43 months, and 32% at 72 months; χ21 = 18.25; P<.001 for trend).
Conclusions Our results suggest that anti-PRP antibody levels and clinical protection
against Hib disease wane over time after Hib vaccination at 2, 3, and 4 months
of age without a booster dose at 2 years of age. The decline in clinical protection
is minimal, however, suggesting that a booster dose of Hib vaccine following
infant vaccination is not essential.
Haemophilus influenzae type b (Hib) is an important
cause of meningitis, septicemia, pneumonia, and epiglottitis in children worldwide.
Many countries have introduced Hib conjugate vaccines into their routine childhood
vaccine schedules and the impact on invasive Hib disease has been uniformly
impressive.1 In general, the schedule chosen
for vaccination against Hib has reflected that already in place for the established
infant vaccines against diphtheria-tetanus-pertussis (DTP) and polio. With
few exceptions, a booster dose of Hib vaccine has been included in the second
year of life. One of the major reasons for this is that vaccine-induced antibody
wanes as children get older,2,3
leading to the concern that this may be accompanied by an increase in susceptibility
to disease. The relationship between antibody to polyribosylribitol phosphate
(PRP, the capsular polysaccharide of Hib) and clinical protection is inferred
from studies of natural immunity, passive immunization, and the plain unconjugated
PRP vaccine. From such studies arose the figures of 0.15 µg/mL as a
correlate of short-term protection and 1.0 µg/mL as a correlate of long-term
protection against invasive Hib disease.4
Experience with the Hib conjugate vaccines has suggested that the serum
anti-PRP antibody concentration may not be as robust a correlate as it was
after PRP polysaccharide vaccination. The first Hib conjugate vaccine, PRP-D
(PRP conjugated to diphtheria toxoid), serves as a good example of this. After
3 doses of this vaccine, typically only 40% of infants achieve antibody concentrations
of 1.0 µg/mL and 70% achieve 0.15 µg/mL.5
Yet, most clinical trials5 as well as widespread
implementation in several populations6,7
suggest clinical protection in more than 90% of vaccinees. The explanation
appears to be that conjugation of PRP to a protein converts it from a T-cell–independent
to a T-cell–dependent antigen with the capacity to induce immunological
memory.8 In the presence of immunological memory,
a rapid and protective increase in antibody concentration may be achieved
in response to exposure to the organism, despite a low background antibody
In the United Kingdom, routine immunization with Hib conjugate vaccines
was introduced in October 1992 at the primary infant DTP schedule of 2, 3,
and 4 months of age. No booster dose of Hib vaccine was included in the schedule.
There was catch-up vaccination for children up to 48 months of age that took
place over the first year of the program: 3 doses of vaccine were given to
infants younger than 1 year with a single dose to older children. A national,
active, prospective survey of Hib vaccine failures was established simultaneously.
The major objective of this survey was to assess the effectiveness of the
Hib vaccine program in the absence of a routine booster dose. Surveillance
was widened in 1995 to cover H influenzae disease
in all children, regardless of vaccination status.
In this article we present the first 6 years of surveillance of vaccine
failures as well as 3 years of surveillance of Hib disease in unvaccinated
children. In addition, we present longitudinal anti-PRP antibody data from
children who were vaccinated with 3 doses of PRP-T (PRP conjugated to tetanus
toxoid) conjugate vaccine during prelicensure immunogenicity studies in Oxford,
England, and have now been followed up to school age. The major objective
of these studies was to evaluate long-term clinical protection against Hib
disease and Hib antibody concentrations following primary Hib vaccination
without a booster dose.
This study was performed under the auspices of the British Paediatric
Surveillance Unit (BPSU) of the Royal College of Paediatrics and Child Health.
The BPSU has a program of active surveillance for selected rare pediatric
conditions in the United Kingdom. Every month more than 90% of pediatricians
routinely report to the BPSU.9
Pediatricians were asked to provide early notification by telephone
of a case of H influenzae disease (b and non-b H influenzae disease were eligible). They were sent a questionnaire
requesting clinical, demographic, and laboratory information. Microbiologists
and consultants in communicable disease control (CCDCs) were informed about
the study and encouraged to contribute to surveillance by notifying the BPSU
about cases and sending isolates of H influenzae
to the national reference laboratory.
For the first phase of the study (October 1, 1992-October 31,1995),
reports were requested of any child with invasive H influenzae disease who had received Hib conjugate vaccine. From November 1, 1995,
the case definition was broadened to include all children with invasive H influenzae disease regardless of vaccination status. Invasive disease was defined as isolation of H influenzae from a normally sterile site or a positive Hib antigen
test combined with a clinical picture compatible with invasive Hib disease.
The dates of all primary immunizations were obtained from the child's
general practitioner or from the district child health immunization computer.
The local microbiologist was asked to send the isolate to the Public Health
Laboratory Service Haemophilus Reference Laboratory (Oxford) where the identity
was verified by standard slide agglutination and polymerase chain reaction
To maximize the reporting of cases, the study was widely publicized
in the medical press prior to its commencement, and pediatricians, microbiologists,
and CCDCs were then updated through quarterly BPSU bulletins, an annual BPSU
report, and publications in the Communicable Disease Report (England and Wales) and the Scottish Centre of Infection
and Environmental Health Report. Reports to other agencies of vaccine
failures were checked against those reported to the study including the 2
pharmaceutical companies supplying the vaccines, the UK Department of Health
Medicines Control Agency, the Public Health Laboratory Service Communicable
Disease Surveillance Centre for England and Wales, and the Scottish Centre
for Infection and Environmental Health.
The specific Hib conjugate vaccines and combinations administered to
children changed over the period of the study. Between 1992 and 1996 the dominant
vaccine was PRP-T (Aventis Pasteur, Berkshire, England) given as a separate
injection, with HbOC (PRP conjugated to mutant diphtheria toxoid, CRM197 [Wyeth,
Berkshire, England]) used for catch-up vaccination of children older than
1 year (1992-1994). From 1996 PRP-T (Aventis Pasteur) has been given as a
combination vaccine (with DTP), and HbOC has also been available for primary
vaccination. Since 1997, another PRP-T vaccine (SmithKline Beecham, Hertfordshire,
England) has also been available for primary vaccination (Debby Webb, Department
of Health, London, England, written communication, August 1999). All children
received whole cell pertussis–containing DTP vaccines. For the purposes
of this analysis all of these conjugates have been considered equivalent.
A true vaccine failure (TVF) was defined as
invasive Hib disease occurring more than 2 weeks after a single dose of vaccine
given to a child when older than 1 year, or more than 1 week after at least
2 doses of vaccine had been given to an infant aged 1 year or younger. Clinical
effectiveness data are presented only for infants given 3 doses of vaccine.
A case of Hib disease occurring after receipt of 1 or 2 doses of vaccine but
before sufficient time had elapsed to be counted as a TVF was defined an apparent vaccine failure. If a strain of H influenzae was isolated from a vaccinated child but not sent to the
reference laboratory for verification, it was defined as a possible vaccine failure.
The recruitment methods have been described previously.11- 13
The first cohort of children was enrolled in an immunogenicity and safety
study of PRP-T vaccine. Briefly, parents of infants born at the John Radcliffe
Hospital in Oxford were approached after the birth of their child and given
information about the study. Written informed consent was obtained and infants
received PRP-T vaccine as a separate injection from DTP at 2, 3, and 4 months
of age. No further doses of Hib vaccine were given. A total of 107 infants
were initially enrolled and serum samples were obtained from as many of these
infants as possible at 2 months of age (90), 5 months of age (105), and 12
months of age (95); the results have been reported previously.11,12
A further sample was obtained at 6 years of age from 59 children (first cohort).
A separate group of 153 children born in 1991 and vaccinated with PRP-T
vaccine at 2, 3, and 4 months of age had samples taken at 43 months of age
(second cohort).13 These children were originally
part of a large PRP-T implementation trial that took place in the Oxford Health
Region in 1991-1992.14 The names of these children
were obtained from the Oxford district child health immunization computer
records and the parents were approached by letter. Those who agreed to participate
were visited at home and written informed consent obtained.
Serum was stored at −20° C until serological tests were performed.
Anti-PRP IgG antibodies were quantified using an enzyme-linked immunosorbent
Vaccine Effectiveness. To derive estimates of vaccine effectiveness of Hib conjugate vaccine,
the observed number of TVFs (after 3 doses) that accrued in infants born between
August 1, 1992, and March 1, 1999, was divided into age bands and compared
with the number of cases expected based on the age-specific rates of invasive
Hib disease. These rates were obtained from a 6-year survey that took place
between 1985 and 1990, prior to the introduction of the Hib vaccine in the
Oxford Health Region.15 The population at risk
in the Oxford Health Region survey was calculated using census data for mid
1988 (Office of National Statistics, London). The follow-up (exposure) time
of vaccinated children began from 5 months of age, 1 week after the average
time at which infants complete the primary course of immunization.16 The exposure times were then calculated using the
national vaccine coverage figures for each year from 1992-1998 (COVER [Coverage
of Vaccination Evaluated Rapidly]data, published quarterly in the Communicable Disease Report CDR Weekly) and the annual national birth
and mortality rates (Office of National Statistics, London). The annual UK
birth cohort varied from 717,000 to 781,000 over the period of the study.
Poisson regression models were used to estimate the relative incidence
rate for each age year between the 2 samples. Vaccine effectiveness was calculated
as 100 × (1 − relative incidence rate) and is quoted together
with 95% confidence intervals (CIs). A comparison between the vaccine effectiveness
in the first year of life with subsequent years was computed by including
an interaction term in the model. All statistical analyses were performed
using STATA statistical software.17
Estimates and CIs of the relative vaccine effectiveness comparing those
vaccinated and unvaccinated were made using standard methods for comparing
Anti-PRP Antibody Concentrations. Anti-PRP antibody concentrations were determined at ages 2, 5, 12, and
72 months from the first cohort and at 43 months from the second cohort. Three
sets of comparisons between antibody concentrations were made: (1) between
month 2 and month 5 (to assess the response to vaccine; (2) between month
5 and month 12 (to assess the early decline following completion of the vaccine
schedule); and (3) between months 12, 43, and 72 (to assess the long-term
decline in antibody levels). Comparisons between anti-PRP antibody concentrations
at ages 2, 5, 12, and 72 months were made accounting for the longitudinal
nature of the data, whereas all comparisons involving the 43-month data treated
the data from other time points as if they were from an independent group.
Antibody concentrations are summarized using geometric mean concentrations
(GMCs), and all statistical analyses were performed after log-transforming
the data. Changes in GMCs between time points are reported as fold rises.
Where antibody levels were undetectable (<0.15 µg/mL), the value
0.08 was substituted. The differences between the time points were tested
using t tests, while the trends in antibody levels
over several time points were assessed using linear regression. Changes in
the proportion below the antibody thresholds of 0.15 and 1.0 µg/mL were
analyzed using the χ2 test for trend. Confidence intervals
for proportions were calculated using the Binomial Exact method.
All studies were approved by the Central Oxford Research Ethics Committee.
During the 6-year, 5-month period between October 1, 1992, and March
1, 1999, 112 TVFs and 58 apparent vaccine failures were reported to the study.
There were 7 possible vaccine failures. In the 3-year, 4-month period between
November 1, 1995, and March 1, 1999, 39 cases of invasive Hib disease were
reported in unvaccinated children.
Of 112 TVFs, there were 96 children who had previously received 3 doses
of Hib vaccine, 9 who had received 2 doses, and 7 who had received 1 dose.
There were 4 deaths among the TVFs, 3 occurring after 3 doses of vaccine (ages
5, 17, and 30 months), and 1 after 1 dose.
During the study period an estimated 4,368,200 infants in the United
Kingdom received 3 doses of vaccine; the vaccine failure rate after 3 doses
was therefore 2.2 per 100,000 vaccinees (95% CI, 1.8-2.7).
Table 1 divides the TVFs
that occurred after 3 doses of vaccine according to age at disease and compares
the observed number of TVFs with the number of cases expected, based on the
child years of exposure and the age-specific attack rate observed in prevaccination
surveillance. Vaccine effectiveness at different ages is estimated. Inclusion
of the 7 possible vaccine failures makes no significant difference to estimates
of vaccine effectiveness.
The vaccine effectiveness declined after the first year (P<.001), although the decrease was small (99.4% in children aged
5-11 months compared with 97.3% in children between 12-71 months of age; relative
disease incidence rate, 4.1 [95% CI, 2.3-7.2]).
Performing a sensitivity analysis where the assumption is made that
surveillance for vaccine failures was considerably incomplete by doubling
the number of TVFs actually notified to the study, the vaccine effectiveness
estimates are only minimally affected (Figure
1). The overall vaccine effectiveness calculated by doubling the
actual number of cases reported becomes 96.4% (95% CI, 95.7%-97.0%).
Subdividing TVF by age of a hypothetical booster dose (given at 12-15
months of age plus 1 week to develop a protective antibody response) reveals
that 79 patients were 12.25 months of age or older and 71 patients were 15.25
months of age or older at the time of disease. Two of the 3 deaths after 3
doses were in children older than the age of a hypothetical booster dose.
Thirty-nine cases of Hib disease occurred in unvaccinated children.
There were 3 deaths. The median age at disease for all Hib patients was 8.7
months (range, 0-164 months). Timely Hib vaccination could have prevented
13 of these cases (including 1 death). The remaining 26 were either too young
to have received a protective course of vaccination, too old to have been
eligible for vaccination (born before 1988), were born outside the United
Kingdom, or had underlying conditions and may not have responded to vaccination.
For the period November 1, 1995, to March 1, 1999, all children younger
than 16 years with invasive Hib disease, regardless of vaccination status,
were eligible for notification to the study. The incidence of Hib disease
in infants younger than 1 year and children younger than 5 years for the 3
full years of surveillance (1996, 1997, and 1998) are shown in Table 2 according to vaccination status (receipt of any number of
doses of Hib vaccine). In 1996 the relative risk of a vaccinated child (any
number of doses) having invasive Hib disease compared with an unvaccinated
child was 0.03 (95% CI, 0.01-0.15); in 1997, 0.05 (95% CI. 0.01-0.16); and
in 1998, 0.04 (95% CI, 0.01-0.20).
The GMC of anti-PRP antibody and the proportions of children with an
anti-PRP antibody concentration below 0.15 µg/mL and 1.0 µg/mL
at different ages are shown in Table 3.
A significant increase in GMC was found between months 2 and 5 (13-fold increase, P<.001) and a significant decrease in the period between
month 5 and month 12 (5-fold decline, P<.001).
However, the decline in GMC between month 12 and month 72 was small and nonsignificant,
both when estimated by comparing paired serum samples from month 12 with month
72 (1.35-fold decline, P = .30) and when considering
the trend between months 12, 43, and 72 (1.09-fold decline per year; 95% CI,
0.98-1.20; P = .13). The proportion with levels below
0.15 µg/mL significantly increased during this period (χ21 test for trend = 18.25, P<.001).
In Figure 1, vaccine effectiveness
is shown in different age groups comparing that estimated by clinical vaccine
failure rates (after 3 doses) with that predicted by the proportions with
anti-PRP antibody concentrations greater than 0.15 µg/mL and greater
than 1.0 µg/mL (after 3 doses).
The introduction of routine Hib vaccination in the United Kingdom has
resulted in a rapid, dramatic, and sustained decline in the incidence of Hib
disease. The incidence in children younger than 5 years in 1998 was 0.6 per
100,000. In contrast, prior to the introduction of vaccination, the incidence
for England and Wales was 31 to 36 per 100,000,15,19
a reduction of 98%. This incidence compares favorably with recent estimates
in other countries. In the United States in 1996 and 1997, the incidence of
Hib disease in those younger than 5 years identified through national surveillance
ranged between 0 and 2.9 per 100,000 in different states (excluding Alaska).
The race-adjusted incidence derived from active laboratory-based surveillance
in selected areas in 1997 was 0.4 per 100,000 in children younger than 5 years
and the decline in incidence between 1989 and 1997 was estimated at 99%.20
Key factors in the success of the UK program are likely to include the
high vaccine coverage achieved as well as a program that included all children
to 48 months of age. The rationale for this strategy was that susceptibility
to Hib disease continues to this age1 and that
pharyngeal carriage of Hib is relatively more common in older preschool children
(summarized by Coen et al21). Inclusion of
this group may therefore have had an important effect on reducing the circulation
of Hib in the population. As in other countries,22,23
pharyngeal carriage rates of Hib have declined in the United Kingdom since
the introduction of Hib vaccination.13 Recent
experience in the Alaskan native population underscores the importance of
reducing Hib transmission through vaccination. A resurgence of invasive Hib
cases in children in 1996 and 1997 was attributed to continuing Hib carriage
unmasked by a change in the vaccination regimen.24
Similar to other investigators,2 we have
shown a decline in anti-PRP antibody concentrations after primary immunization.
In our cohort, the fall in antibody between 5 and 12 months of age was statistically
significant and thereafter, although the decline in GMC was not statistically
significant, the trend in proportions of children with anti-PRP concentrations
below 0.15 µg/mL (a putative protective threshold) was also significant.
This might suggest an increasing susceptibility to invasive Hib disease as
children in the United Kingdom get older. Indeed, in the absence of surveillance
data, it would seem prudent to recommend a second year booster dose to prevent
this apparent decline in protection. However, we have had the unique opportunity
to interpret these serological data in the light of clinical vaccine failure
cases. Vaccine protection, as assessed through clinical failures, remains
high until 6 years of age. Although there is a statistically significant decline
after the first year of life, it is small in real terms with the absolute
estimates of vaccine effectiveness and their 95% lower confidence limits remaining
very high up to 4 years of age. Even if underreporting is assumed, the estimates
of vaccine effectiveness remain high.
A comparison between predicted efficacy from immunogenicity data and
estimated efficacy from surveillance data suggests that a threshold of 0.15
µg/mL correlates much better than a threshold of 1.0 µg/mL, although
even a level of 0.15 µg/mL clearly underestimates effectiveness. Why
do antibody concentrations after receipt of conjugate vaccines not correlate
well with clinical protection? The antibody thresholds of 0.15 and 1.0 µg/mL
were derived from unvaccinated populations and those who received passive
immunization (0.15 µg/mL) and populations who received the PRP capsular
polysaccharide vaccine (1.0 µg/mL).4
In contrast to this vaccine, the conjugate Hib vaccines are capable of inducing
immunological memory in recipients;8 thus,
even those individuals with low antibody concentrations may still be protected
against disease. Immunological memory has been demonstrated using the UK schedule.25 Other laboratory markers of immunological memory,
such as antibody avidity, may now prove more useful than specific antibody
concentration measurement in the context of conjugate vaccines.26
This argument has recently been advanced to support the use of acellular pertussis/Hib
conjugate vaccine combinations despite the lower Hib antibody concentrations
frequently encountered with such combinations.27
Conjugate vaccines also have an impact on carriage of Hib and thereby
reduce Hib circulation and provide benefit through reduced transmission of
Hib in the population (herd immunity). Thus, the unvaccinated, partially vaccinated,
or vaccine failures may avoid coming in contact with Hib. This is demonstrated
in this article by the 10-fold reduction in Hib disease seen in unvaccinated
infants younger than 1 year when compared with their rate in the prevaccine
era. Thus, both immunological memory and herd immunity are likely to be contributing
to the successful impact of Hib conjugate vaccines in the United Kingdom.
If a booster dose of Hib vaccine were to be incorporated into the UK
schedule, it might logically be given at the same time as the measles-mumps-rubella
vaccine, ie, between 12 and 15 months of age. If such a program had been in
place, with an uptake of 100% and vaccine effectiveness of 100%, then 71 to
79 Hib cases including 40 to 44 cases of meningitis and 2 deaths may have
been avoided. However, approximately 4.2 million booster doses would need
to have been administered in the United Kingdom to do so–equivalent
to 100,000 doses for each case of Hib meningitis prevented. This further assumes
that a booster dose would have been effective (25% were known to have had
clinical risk factors and/or low immunoglobulin concentrations that may have
impaired response to vaccination28). A formal
cost-benefit analysis is beyond the scope of this article, but would be of
interest to countries that currently recommend a booster dose as they reevaluate
their immunization schedules.
The overall impact of routine immunization in the United Kingdom with
a 3-dose primary schedule without a fourth dose is similar or greater in magnitude
to Iceland,7 the United States,20
Canada,29 the Netherlands,30
and Australia,31 which have routinely included
a fourth dose. The impact of the UK program is also similar to that witnessed
in those Scandinavian countries that also use only 3 doses of Hib vaccine
but administer 2 doses in the first year and the third dose in the second
year of life.32
The demonstration that a booster dose in the second year of life is
not essential has significant implications for countries yet to initiate routine
Hib vaccination. In particular, in much of the developing world, Hib is an
important pathogen,33 yet there are few developing
countries that can afford the relatively expensive Hib conjugates.34 The decision to introduce this vaccine would be made
easier if fewer doses were required. Hib conjugate vaccines have recently
been introduced in Chile and The Gambia using 3-dose schedules,35,36
and the early experience in The Gambia is promising.37
Encouraging recent studies in Chile suggest further that 2 doses may suffice.38 Long-term follow-up of persons with Hib disease in
nonindustrialized countries will be critical in extrapolating the need for
a booster dose to all populations.