For many years Haemophilus influenzae type b (Hib) has been recognized as an important cause of serious bacterial infections in young children in industrialized countries. The recent introduction of Hib conjugate vaccines into those countries has virtually eliminated invasive Hib disease, yet we still know remarkably little about Hib disease in developing countries. Studies of Hib meningitis from African countries have shown the incidence of the disease to be 250 to 300 cases per 100 000 children per year in the first year of life, falling off thereafter. The incidence is higher than in industrialized countries, and the disease occurs in younger children. Furthermore the outcome is worse, with reported mortality rates in African studies of 20 to 50%1-3 and serious neurologic sequelae occurring in a substantial proportion of survivors. In most developing countries overall mortality rates are likely to be underestimated to a degree determined by the number of cases occurring in areas without reasonable medical facilities, where the mortality is likely to approach 100%. Although there is no doubt that Hib meningitis is a major problem in Africa, it is probably much less common than Hib pneumonia.
In most cases of pediatric pneumonia, it is not possible to determine the causative organism. Blood cultures may be positive in up to 20% of cases,4-6 but this is not always an accurate guide to the true etiology.7 The technique of percutaneous lung aspiration can improve the detection rate to up to 50%,7 but the etiology of many cases still remains in doubt. The World Bank estimates that every year 2.7 million children <5 years of age die of pneumonia, most in developing countries, many of them in Africa. To gauge the potential impact of Hib vaccination, it is necessary to estimate the proportion of such deaths that is caused by Hib pneumonia. Until now this has been done by extrapolating from published studies of the etiology of pneumonia, but the use of such data for this purpose is logically flawed.
A number of the pneumonia studies used to provide these estimates for Africa have been performed in The Gambia, a small developing country in West Africa.5, 6, 8-10 Most of those studies were performed at the Medical Research Council Hospital located in Fajara, an urban setting on the coast about 15 km from Banjul, surrounded by the wealthiest residential area in the country. Although many of the communities using the hospital are poor, all have access to primary health care from a number of sources in the community, and many children with bacterial pneumonia are treated successfully at this level. Thus the population of children with pneumonia reaching the hospital ward, although containing the more severe of the treatable bacterial causes, will be biased toward those causes of severe lower respiratory tract infections that do not respond to simple antibiotics used in the community, particularly respiratory syncytial virus (RSV) as well as other viruses and bacteria such as Staphylococcus aureus and Klebsiella pneumoniae.
Use of these data to estimate the causes of death from pneumonia introduces another bias in the same direction. Most of the 2.7 million or so children <5 years of age who die from pneumonia every year do so because of lack of access to simple antibiotics. If a typical African hospital were to admit 100 infants with severe lower respiratory tract infection, 50 caused by RSV and 50 caused by a bacterial pathogen, the overall mortality (probably about 10%) will be contributed to by both groups but probably more because of the bacteria than the RSV (perhaps 7 cases vs. 3 cases). However, if no antibiotics were available in that hospital, the mortality rate in the bacterial group would rise steeply to around 30%.11 Thus if we had extrapolated from the hospital data based on treated children to estimate the causes of (untreated) pneumonia mortality in the community, we would have concluded that 70% are caused by bacteria and 30% by RSV, whereas the true figure, based on the likely untreated mortality, is closer to 90% bacterial and 10% RSV. Again the use of hospital-based data results in overestimation of the nonbacterial causes and underestimation of the bacterial causes.
In The Gambia hospital-based etiology studies in which blood culture and percutaneous lung aspiration were used had lead to the conclusion that 26% of severe pneumonia was caused by S. pneumoniae and 7% by Hib.6 The conduct of a field trial of a Hib conjugate vaccine offered the opportunity to examine the estimates for Hib pneumonia from a different perspective.
THE GAMBIAN HIB VACCINE TRIAL
Details of the trial have been published elsewhere.13 Briefly 42 800 Gambian infants presenting for their first diphtheria-tetanus-pertussis (DTP) vaccine at 20 health centers in the western region of The Gambia between March, 1993, and October, 1995, were randomized to receive DTP mixed with Hib polysaccharide-tetanus toxoid conjugate vaccine [PRP-T (Act-Hib®), Pasteur Mérieux Connaught, France] or DTP mixed with placebo at the times when they presented for the infant DTP series. This was scheduled to be at 2, 3 and 4 months of age, but in practice the median ages at which children in the study received the vaccines were 11, 18 and 24 weeks, with 17% of children failing to receive all 3 doses. Cases of meningitis, pneumonia and other possible invasive Hib diseases were investigated by blood and cerebrospinal fluid culture at the 3 hospitals in the study area. Chest radiography and percutaneous lung aspiration were performed at the 2 hospitals with radiology facilities.
When surveillance for the study was completed in February, 1996, 50 cases of invasive Hib disease had been detected among children enrolled in the study: 17 pneumonia; 28 meningitis; and 5 other Hib disease cases. Among children regarded as having effectively received at least 2 doses of vaccine (i.e. 14 or more days had elapsed between the second dose and the day of admission), the efficacy of the vaccine for the prevention of bacteriologically proved Hib pneumonia was 100% [controls 10, vaccinees 0; 95% confidence interval (95% CI), 55% to 100%] and the efficacy for prevention of all invasive Hib disease was 93% (controls 29, vaccinees 2; 95% CI, 73% to 99%). After 2 or 3 doses the vaccine appears to have prevented 15 cases of meningitis and 10 cases of pneumonia. The number of pneumonia cases prevented seems small, but most cases of pneumonia were not etiologically confirmed.
To determine the true impact of the vaccine on overall pneumonia rates and therefore by implication to estimate the proportion of pneumonia that is attributable to Hib, an analysis was performed of all episodes of acute lower respiratory tract infection detected among study children. Of the 1793 acute lower respiratory tract infection episodes that were investigated by the study physicians, 864 were in the vaccine group and 929 were in the placebo group, suggesting that PRP-T prevented 65 episodes due to Hib. A chest radiograph was obtained in 1269 acute lower respiratory tract infection episodes (71%). At the end of the study these were read in a blinded manner by the pediatricians involved in the study using a simple recording system that categorized the films as showing possible or definite consolidation. Lobar consolidation and pleural effusion were recorded separately.
Of the 1269 chest radiographs read in this manner, 449 showed definite radiologic consolidation (198 in the vaccine group, 251 in the placebo group), of which 201 were lobar consolidation or consolidation with effusion (vaccine, 86; placebo, 115). Thus, it appears that PRP-T prevented 53 episodes of definite radiologic pneumonia (21%) and 29 episodes of lobar pneumonia or pneumonia with effusion (26%). Indeed this analysis certainly must underestimate the actual number of pneumonia episodes prevented, because only a proportion (perhaps one-half) of episodes of pneumonia in the study children would have been investigated and prompted a chest radiograph. Many episodes would have been missed by the surveillance system, and because of logistic or technical problems only 71% of those detected prompted a chest radiograph.
Despite the limitations these data suggest that 21 to 26% of episodes of severe pneumonia in children >12 weeks of age in this community were caused by Hib. This appears to be a much better estimate of the burden of Hib disease in this community than had been obtained from hospital-based etiology studies, but for two reasons this still may be an underestimate: (1) the estimate assumes that the vaccine is 100% efficacious. Although our point estimate of efficacy for Hib pneumonia after two doses was 100%, the 95% confidence interval extended down to 55%. Our estimate of the true contribution of Hib will be an underestimate by the degree to which the true vaccine efficacy is <100%; (2) it ignores the effect of herd immunity, which will reduce Hib disease rates in the control group and thus make the difference between the groups appear smaller.
Herd immunity, or the ability of a vaccine to have a greater impact on disease than would be anticipated based on knowledge of the vaccine efficacy and coverage, has been well-documented with the use of Hib conjugate vaccines in industrialized countries and has been shown to be the result of reduced oropharyngeal carriage of the organism in vaccine recipients. Many doubted that this would occur in African children, in whom carriage rates are higher and probably quantitatively heavier and the intensity of transmission greater. The Gambian study provided the opportunity to examine the herd effect in two ways: protection offered by the vaccine against oropharyngeal carriage; and rates of carriage and disease in the unvaccinated controls.
Details of the carriage study are published elsewhere.13 Briefly 4000 children 12 to 24 months of age underwent oropharyngeal swabbing for selective culture of Hib; 1000 children were studied in each of 4 years spanning the study. Children studied in the final 2 years were study participants, and from that group it was determined that the vaccine given in infancy reduced the likelihood of a child's carrying Hib in the second year of life by 60% (95% CI, 44% to 72%). The rate of Hib carriage in the unvaccinated children fell steadily during the 3 years of the study, suggesting the herd effect that would be expected given the effect of the vaccine on carriage. However, an unexpectedly low carriage rate in the year before the study makes it difficult to draw conclusions.
The actual herd effect could be observed by analyzing the rate of invasive Hib disease in the control group during the trial, when ∼45% of the infant population was receiving the vaccine. The rate of detection of invasive Hib disease appeared to fall during the course of the study, but month-to-month variability in Hib disease rates makes it again difficult to draw conclusions. Despite more intense surveillance than was provided in the earlier surveillance studies in the same area, the rate of Hib disease in the control group was slightly lower than was found in the surveillance studies, suggesting but not conclusively demonstrating a herd effect. The incidence of invasive Hib disease in control infants 2 to 12 months of age in the 1993 to 1995 vaccine trial was 236 per 100 000 child years (95% CI, 160 to 312) compared with 274 per 100 000 child years during the period 1990 to 199314 and 297 per 100 000 child years from 1986 to 1988.2 It is possible that an even larger herd effect was obscured by the more intense surveillance associated with the main trial.
The Gambian Hib trial provided an excellent opportunity to understand better the contribution of Hib to the pneumonia burden of African infants. The data suggest that ∼21% of episodes of severe pneumonia in infants >2 months of age are caused by Hib. The confidence intervals around the estimate are wide and the exact rate is uncertain, yet examination of the likely biases in the estimation of this figure suggests that if anything, the real contribution of Hib to postneonatal mortality in this community may be higher.
The figure of 21% is more than double the estimates of the contribution of Hib based on earlier etiology studies from The Gambia,10 confirming that extrapolation from hospital-based etiology studies does indeed underestimate the contribution to overall pneumonia morbidity, and presumably mortality, caused by pathogens such as H. influenzae and S. pneumoniae that respond to simple antibiotics. The same biases are likely to result in the systematic underestimation of the potential impact of pneumococcal vaccines. The true impact of a pathogen can be gauged only by eliminating the pathogen with a vaccine, yet for the reasons outlined above, even a double blind vaccine trial will underestimate the impact because of the herd effect and less than perfect vaccine efficacy.
Estimates of the global burden of various diseases and pathogens are about to be published. Of necessity these are rough estimates intended to help health planners and researchers set priorities. Transparency is essential for the presentation of such estimates, so that educated readers can judge the likely biases themselves. This is particularly true for complex calculations such as estimates of pneumonia morbidity and mortality by etiology. In the case of Hib, estimates based on observational studies have been tested and improved by a specific vaccine trial. The same process may be required to determine the true burden of other pathogens the true role of which is unclear in disease, such as Helicobacter pylori or Plasmodium falciparum. Yet most interesting of all may be the impact of the new generation of conjugate vaccines against S. pneumoniae. If the experience with Hib is repeated with S. pneumoniae, current estimates of the disease burden resulting from that pathogen may prove to be embarrassingly low.
The relatively modest herd immunity effect seen with the vaccination of ∼45% of infants in the population in this trial contrasts with the more dramatic effect seen in the United States. The measured herd immunity probably is an underestimate because of the more intense surveillance used for the trial. Nevertheless evaluation of the true magnitude of the herd effect will await postintroduction studies in a developing country. Such a study is currently under way in The Gambia.
The finding of a high degree of efficacy for the prevention of meningitis and pneumonia caused by Hib in African infants was expected by many. However, the significant impact on overall rates of severe pneumonia and the demonstration of the true contribution of Hib to the childhood pneumonia burden were unexpected and very welcome findings. In this setting a vaccine trial proved to be a very powerful instrument for the estimation of disease burden attributable to a pathogen.
The Gambian Hib trial was conducted by the Medical Research Council (UK) with support from the United States Agency for International Development, the United Nations Children's Fund, the World Health Organization and the Children's Vaccine Initiative. We thank Shabbar Jaffar for help with the statistics and Professor Brian Greenwood and the other members of the Gambian Hib trial team for permission to publish the previously unpublished data in this paper.
1. Bijlmer HA. Epidemiology of Haemophilus influenzae
invasive disease in developing countries
and intervention strategies. In: Ellis RW, Granoff DM, eds. Development and uses of Haemophilus
b conjugate vaccines. New York: Dekker, 1994:247-64.
2. Bijlmer HA, van Alphen L, Greenwood BM, et al. The epidemiology of Haemophilus influenzae meningitis
in children under 5 years of age in The Gambia
, West Africa. J Infect Dis 1990;161:1210-5.
3. Funkhouser A, Steinhoff MC, Ward J. Haemophilus influenzae
disease and immunization in developing countries
. Rev Infect Dis 1991;13(Suppl 6):S542-54.
4. Shann F, Gratten M, Germer S, Linnemann V, Hazlett D, Payne R. Aetiology of pneumonia
in children in Goroka Hospital, Papua New Guinea. Lancet 1984;2:537-41.
5. Wall RA, Corrah PT, Mabey DCW, Greenwood BM. The etiology of lobar pneumonia
in The Gambia
. Bull WHO 1986;64:553-8.
6. Adegbola RA, Falade AG, Sam BE, et al. The etiology of pneumonia
in malnourished and well-nourished Gambian children. Pediatr Infect Dis J 1994;13:975-82.
7. Falade AG, Mulholland EK, Adegbola RA, Greenwood BM. Bacterial isolates from blood and lung aspirate cultures in Gambian children with lobar pneumonia
. Ann Trop Paediatr 1997;17:315-9.
8. Forgie IM, O'Neill KP, Lloyd-Evans N, et al. Etiology of acute lower respiratory tract infections in Gambian children: I. Acute lower respiratory tract infections in infants presenting at the hospital. Pediatr Infect Dis J 1991;10:33-41.
9. Forgie IM, O'Neill KP, Lloyd-Evans, et al. Etiology of acute lower respiratory tract infections in Gambian children: II. Acute lower respiratory tract infection in children ages one to nine years presenting at the hospital. Pediatr Infect Dis J 1991;10:42-7.
10. Greenwood BM. Epidemiology of acute lower respiratory tract infections, especially those due to Haemophilus influenzae
type b, in The Gambia
, West Africa. J Infect Dis 1992;165(Suppl 1):S26-8.
11. Greengard W, Raycraft WB, Motel WG. Effect of chemotherapy on pneumonia
in infants under one year of age. Am J Dis Child 1943;62:730-42.
12. Mulholland EK, Hilton S, Adegbola RA, et al. Randomized trial of Haemophilus influenzae
type b-tetanus protein conjugate vaccine
for prevention of pneumonia
in Gambian infants. Lancet 1997;349:1191-7.
13. Adegbola RA, Mulholland EK, Secko O, et al. Vaccination with Haemophilus influenzae
type b conjugate vaccine
reduces oropharyngeal carriage of Haemophilus influenzae
type b among Gambian children. J Infect Dis 1998;177:1758-61.
14. Adegbola RA, Mulholland EK, Falade AG, et al. Haemophilus influenzae
type b disease in the Western Region of The Gambia
: background surveillance for a vaccine
efficacy trial. Ann Trop Paediatr 1996;16:103-11.
FIRST INTERNATIONAL CONFERENCE ON HAEMOPHILUS INFLUENZAE TYPE b INFECTION IN ASIA
The Editors thank the Association pur l'Aide à la Médicine Préventive, the Foundation Mérieux, and the World Health Organization for supporting publication of these proceedsings, and Jennifer Wells for her editorial assistance.