Annually an estimated 350 000 to 700 000 children in developing countries die from Haemophilus influenzae type b (Hib) infections.1, 2 In contrast to the situation in industrialized countries, the majority of Hib disease in developing countries manifests as pneumonia.3, 4 The ability of Hib conjugate vaccines to prevent culture-confirmed Hib invasive disease is well-established.5-10 Randomized trials of Hib vaccine efficacy in Santiago, Chile, and The Gambia have shown that Hib vaccines protect against culture-proved Hib pneumonia.5, 6 The Gambian trial also showed a 21% overall reduction in radiographically confirmed pneumonia and a 25% reduction in lobar pneumonia or pneumonia with effusion,6 showing that Hib vaccination provides significant protection against pneumonia that is negative by blood culture or lung aspirate.
As the only other nonindustrialized area (besides The Gambia) where Hib conjugate vaccines have been evaluated in a large scale, randomized study, Santiago, Chile, provides an unique opportunity to assess the efficacy of Hib conjugate vaccine for prevention of radiographically confirmed pneumonia. We performed an analysis of the effectiveness of two or three doses of Hib conjugate vaccine for prevention of pneumonia requiring hospitalization and in particular, pneumonia with consolidation or pleural effusion or other indicators of likely bacterial infection.
Prospective evaluation of PRP-T vaccine in metropolitan Santiago. A large scale, randomized, prospective evaluation of the effectiveness of PRP-T vaccine for prevention of invasive Hib disease was carried out in the 6 health administrative areas of Santiago, Chile, between November, 1992, and March, 1995.5 The methods have been described in detail elsewhere.5 Briefly 36 of the 71 urban health centers in Santiago where children receive vaccines through the Chilean Expanded Program on Immunization were randomized to administer PRP-T vaccine mixed in the same syringe with diphtheria-tetanus toxoids-pertussis (DTP) for 1 year; infants immunized in the remaining 35 health centers received DTP alone. Surveillance data demonstrated a similar background risk of invasive Hib disease in these groups.
During the PRP-T vaccine evaluation, nurses from Centro para Vacunas en Desarrollo-Chile created a database of all vaccinations administered that included the infant's name, date of vaccination, health center and type of vaccine received. For 30 months nurses from the center conducted laboratory-based surveillance to identify patients with invasive Hib disease. This study demonstrated significant protection against invasive Hib disease (92% efficacy) and bacteremic pneumonia/empyema (80%).5
Surveillance of hospitalizations for pneumonia. For logistic and economic reasons this analysis of pneumonia hospitalizations was restricted to infants residing in 2 of the 5 administrative areas (Area Norte and Area Occidente). In these 2 areas 10 274 infants received at least 2 doses of DTP, and 11 146 received at least 2 doses of PRP-T vaccine with DTP during the vaccine trial and provided 53 900 child years of observation for this study. Almost all resident infants and children with an illness severe enough to require hospitalization are served by 3 public hospitals in these administrative areas (Roberto del Rio Children's Hospital in Area Norte; Felix Bulnes Hospital and San Juan de Dios Hospital in Area Occidente).
Hospitalizations for pneumonia were identified by retrospective review of discharge diagnoses. In two of the surveillance hospitals computerized records of discharge diagnoses were available for some or all of the years. Computerized records were searched by ICD-9 code with the following codes: 480.9, viral pneumonia, agent unspecified; 481, pneumococcal pneumonia; 482.x, bacterial pneumonia, other; 485.x, bronchopneumonia, organism unspecified; and 486, pneumonia, unspecified. For the third hospital, Felix Bulnes, and for Roberto del Rio in 1993, hospitalizations for pneumonia were identified by a research physician or nurse who went through the discharge sheets to identify patients with discharge diagnoses that included bronchopneumonia, pleuropneumonia, pneumonia or acute lower respiratory infection. The pneumonia and vaccination databases were then merged to identify those children with pneumonia who were vaccinated during the Hib trial. Children who were in the database of pneumonia admissions but who were not in the database of infants vaccinated during the PRP-T evaluation were either nonresidents of the surveillance area or unvaccinated residents.
Finally 608 episodes of pneumonia requiring hospitalization were identified among children 4 to 23 months old who had received at least 2 doses of DTP or DTP/Hib vaccines.
Clinical and radiologic data collection. Of the 32 494 discharge diagnoses screened, 7716 (24%) were diagnoses compatible with acute lower respiratory tract illness. Of these 7716, 1339 (17%) had a birth date consistent with possible participation in the Hib vaccine evaluation. In participants in the Hib vaccine evaluation 839 episodes were identified, and the medical records were found for 759 (90%) of these episodes. One study physician in each hospital abstracted from the medical records information on the clinical and radiologic features associated with each admission. Information collected included the admitting physician's findings from physical examination (including auscultation), the results of blood or pleural fluid cultures, laboratory test results and the child's condition at discharge. Of the 759 records abstracted, 743 (98%) indicated that a chest radiograph was performed. In each hospital chest radiographs are routinely interpreted by a radiologist.
To characterize the radiologic findings of each episode, study physicians abstracted information from the radiologist's report. If no report could be found the radiograph was interpreted by a radiologist who was blinded to the child's vaccination status. For each admission study physicians noted the distribution of the infiltrates, the type of infiltrate and the presence or absence of hyperinflation, atelectasis and pleural effusion. Each episode was categorized into one of four radiographic patterns: alveolar consolidation only; interstitial infiltrates only without hyperinflation; interstitial infiltrates only with hyperinflation; or mixed (interstitial and alveolar).
Analytic strategy. Analyses included infants who had received two or three doses of either DTP alone or PRP-T vaccine with DTP before their date of admission. Incidence density rates were calculated for each vaccine group by dividing the estimated number of person years of observation in each age group by the number of episodes in that age group. Point estimates of vaccine efficacy were calculated as (1 − RR), where RR equals the incidence rate ratio (incidence among vaccinated divided by incidence among unvaccinated). Pearson's chi square test was used for comparison of incidence rates and categorical variables; the Kruskal-Wallis test was used for comparison of continuous variables that were not normally distributed. P ≤ 0.05 and 95% confidence intervals that did not include 0% efficacy were considered statistically significant results.
The incidence of pneumonia with consolidation or pleural effusion in infants and children 4 to 23 months old was reduced by 22% (P = 0.12) in the group who received PRP-T Hib conjugate vaccine (Table 1). By contrast no differences were observed in the incidence of pneumonia associated with an interstitial pattern or a mixed pattern (alveolar consolidation and interstitial pattern) (3.8 vs. 3.9 and 8.1 vs. 8.0 cases/1000 child years of observation in those who received DTP alone vs. those who received PRP-T/DTP, P = 0.93 and 0.90, respectively). Age-specific differences in the incidence rates were also observed. The incidence rate was greater among infants 4 to 11 months old than among children 12 to 23 months old.
A major aim of this analysis was to determine as accurately as possible how many cases of pneumonia could be prevented by Hib vaccination. As such we aimed to develop a definition of likely bacterial pneumonia that was as sensitive as possible but without compromising specificity. Although radiographic evidence of alveolar consolidation or pleural effusion is a useful indicator of a pneumonia episode that is likely to be bacterial, not all bacterial episodes are characterized by this pattern, and a small proportion of episodes did not have radiographic information available (7%). Nearly one-half (47%) of all episodes were characterized as having evidence of both interstitial infiltrates and alveolar consolidations. In an effort to identify the patients who were likely to have bacterial pneumonia but did not have a radiographic pattern showing exclusively alveolar consolidation, we investigated the association between episodes with radiographic evidence of consolidation and various auscultation and laboratory findings at admission. Episodes of pneumonia associated with bronchial breath sounds were significantly more likely to have a radiographic pattern of alveolar consolidation than episodes associated with rales or with wheezing (43% vs. 25 and 17%, respectively; P < 0.001). The median erythrocyte sedimentation rate (ESR) was significantly higher among pneumonia cases with radiographic evidence of consolidation than among episodes with interstitial or mixed radiographic patterns (38 mm/h vs. 23 and 30 mm/h, respectively; P = 0.002). Consistent with these observations PRP-T provided significant protection against pneumonia associated with bronchial breath sounds or with an ESR ≥ 40 mm/h, but not against pneumonia associated with wheezing or rales (Table 2).
On the basis of these observations we expanded our definition of likely bacterial pneumonia beyond the findings of radiographic evidence of consolidation or effusion. The aim of these analyses was to determine a definition of likely bacterial pneumonia that was sensitive enough to show the full impact of Hib vaccination for prevention of pneumonia by identifying pneumonia cases that may not have been identified by the strict definition of exclusively alveolar consolidation or pleural effusion findings on radiograph. Table 3 shows how addition of bronchial breath sounds and an ESR ≥ 40 mm/h identifies additional cases of likely bacterial pneumonia (increasing sensitivity) without sacrificing vaccine efficacy (an indicator of specificity). Adding ESR ≥ 40 mm/h or bronchial breath sounds to the case definition increases the incidence of pneumonia cases prevented by 63 to 72% (from 1.1 to 1.8 and 1.9 cases prevented per 1000 vaccinees, respectively). The number of cases prevented per 1000 vaccinees more than doubles with the inclusion of both bronchial breath sounds and an ESR ≥ 40 mm/h. The estimated vaccine efficacy varies very little, however, from 28% according to a definition of only alveolar consolidation or pleural effusion to 23% when an ESR ≥ 40 mm/h is used.
This randomized study shows the substantial burden of pneumonia preventable by Hib conjugate vaccination. As the only other nonindustrialized site where a randomized controlled study of Hib conjugate vaccine has been conducted, this study provided an opportunity to build on the results of the Gambian Hib vaccine trial. The results are consistent with those from The Gambia where Mulholland et al. showed that ≥2 doses of PRP-T vaccine reduced lobar pneumonia or pneumonia with effusion by 25%.6 Age-specific efficacy estimates and incidence rates were not included in the analysis from The Gambia. By estimating the incidence of pneumonia requiring hospitalization and by stratifying on age, we were able to calculate the number of pneumonia cases prevented per 1000 vaccinees and show that the impact of Hib vaccination was primarily observed in infants <12 months old, the age group with the highest risk of pneumonia. In this regard our study corroborates and expands the findings from The Gambia.
This study has shown the effect of Hib conjugate vaccination for prevention of pneumonia cases that are negative by standard culture techniques. For example in the cohorts studied here, there were 178 cases of pneumonia with consolidation or effusion but only 2 yielded H. influenzae-positive cultures from blood or pleural fluid (1 patient had been vaccinated with PRP-T). Previously we showed that the PRP-T vaccine in Santiago prevented ∼47 cases of culture-confirmed invasive Hib disease per 100 000 vaccinated children <24 months old.5 In this study the number of hospitalized, nonbacteremic pneumonia cases prevented in children <24 months old was >5 times greater than the number of meningitis cases prevented (250 vs. 47 per 100 000 infants). In the Metropolitan region of Santiago, where there are ∼240 000 children <24 months old, this means that Hib conjugate vaccination potentially prevents ∼600 nonbacteremic pneumonia cases and 112 meningitis cases among children <24 months old each year.
Demonstrating the impact of Hib vaccination is complicated because the most common manifestation of Hib disease, pneumonia, is the most difficult in which to establish an etiologic diagnosis, whereas meningitis, although less common, more often yields a bacterial agent.3, 11, 12 Consequently efforts to document the impact of Hib vaccination often focus on meningitis as a primary outcome. Several studies have demonstrated the effectiveness of routine Hib vaccination for control of Hib meningitis and other culture-confirmed invasive Hib disease outcomes.5, 6, 8, 9, 13, 14 This study shows that, in terms of cases prevented, the impact of Hib vaccination on pneumonia incidence in developing countries will likely be severalfold greater than the impact on meningitis. In this sense our study represents an example of the way in which vaccines can be used as 'probes' to unmask the burden of preventable disease that may be missed by culture-based methods. Decision makers in developing countries should consider this point when estimating the potential effect of Hib vaccination on child health, and they may want to consider the use of "vaccine probe" studies as a way to determine the local burden of vaccine-preventable pneumonia.
In general H. influenzae is usually suspected in patients with lobar pneumonia or other radiographic signs of consolidation, and less often suspected, in patients with a pattern of interstitial or diffuse infiltrates.6, 15 Although radiologic evidence of alveolar consolidation is a highly specific definition of pneumonia that increases the likelihood that the episode has a bacterial etiology, many additional bacterial pneumonias will not meet this strict radiologic criterion. In this study bronchial breath sounds and erythrocyte sedimentation rate were useful as additional indicators of likely bacterial pneumonia. Nevertheless it is important to bear in mind that this study was not designed to assess the utility of bronchial breath sounds or erythrocyte sedimentation rates for predicting bacterial pneumonia in a clinical setting, and these results should not be extrapolated beyond this setting. In future studies where defining the complete burden of preventable disease is a goal, investigators could use a specific case definition for initial analyses and then include additional criteria that do not substantially reduce vaccine efficacy in the case definition.
In our study surveillance was designed to identify a group of pneumonia patients who were most likely to include patients with bacterial pneumonia. It may have missed some patients who were discharged with another code that was less specific for bacterial pneumonia. Moreover outpatients with pneumonia were not included in the surveillance. Hence this analysis is a conservative estimate of the number of cases prevented by Hib vaccination.
Because of the low yield of bacterial cultures from patients with pneumonia, alternative methods to determine the etiology of pneumonia have been proposed, mainly based on nonculture methods such as antigen or nucleic acid detection and serology.16-20 Their specificity, however, may not be sufficient for use in a vaccine trial. The results of this study and those from The Gambia show the utility of using vaccine as a "probe" to estimate the contribution of any given etiologic agent to all pneumonia.16, 21 This experience should be useful for future evaluations of pneumococcal conjugate vaccines, which are expected to have an impact on severe pneumonia.22, 23
The study was supported by Cooperative Research Agreement U01-AI-35948 and Research Contract NO1-AI-45251 from the US National Institute for Allergy and Infectious Diseases. We thank Drs. Kate O'Brien, Anne Schuchat and Ben Schwartz, CDC, and Dr. Alfredo Avendaño, Roberto del Rio Children's Hospital, Santiago, for comments on earlier drafts of the manuscript, and Dr. Kim Mulholland, WHO, for his advice on the study design.
1. WHO. Global Program for Vaccines and Immunizations: The WHO position paper on Haemophilus influenzae
type b conjugate vaccines. Wkly Epidemiol Rec 1998;73:64-7.
2. Levine OS, Schwartz B, Pierce NF, Kane M. Development, evaluation and implementation of Hib vaccines for young children in developing countries: current status and priority actions. Pediatr Infect Dis J 1998;17:S95-113.
3. Greenwood B. 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.
4. Lehmann D. Epidemiology
of acute respiratory tract infections, especially those due to Haemophilus influenzae,
in Papua New Guinean children. J Infect Dis 1992;165(Suppl 1):S20-5.
5. Lagos R, Horowitz I, Toro J, et al. Large-scale, post-licensure, selective vaccination of Chilean infants with PRP-T conjugate vaccine: practicality and effectiveness in preventing invasive Haemophilus influenzae
type b infections. Pediatr Infect Dis J 1996;15:216-22.
6. Mulholland K, Hilton S, Adegbola R, et al. Randomized trial of Haemophilus influenzae
type b tetanus protein conjugate for prevention of pneumonia and meningitis in Gambian infants. Lancet 1997;349:1191-7.
7. Black SB, Shinefield HR, Fireman B, Hiatt R, Polen M, Vittinghoff E. Efficacy in infancy of oligosaccharide conjugate Haemophilus influenzae
type b (HbOC) vaccine in a United States population of 61 080 children. Pediatr Infect Dis J 1991;10:97-104.
8. Booy R, Hodgson S, Carpenter L, et al. Efficacy of Haemophilus influenzae
type b conjugate vaccine PRP-T. Lancet 1994;344:362-6.
9. Santosham M, Wolff M, Reid R, Hohenboken M, et al. The efficacy in Navajo infants of a conjugate vaccine consisting of Haemophilus influenzae
type b polysaccharide and Neisseria meningitidis
outer-membrane protein complex. N Engl J Med 1991;324:1767-72.
10. Peltola H, Kilpi T, Anttila M. Rapid disappearance of Haemophilus influenzae
type b meningitis after routine childhood immunisation with conjugate vaccines. Lancet 1992;340:592-4.
11. Levine OS, Schwartz B. The rationale for population-based surveillance for Haemophilus influenzae
type b meningitis. Pediatr Infect Dis J 1998;17(Suppl):S195-8.
12. Lehmann D. Epidemiology
of acute respiratory tract infections, especially those due to Haemophilus influenzae,
in Papua New Guinean children. J Infect Dis 1992;165(Suppl 1):S20-5.
13. Black SB, Shinefield HR, Hiatt RA, Fireman BH. Efficacy of Haemophilus influenzae
type b capsular polysaccharide vaccine. Pediatr Infect Dis J 1988;7:149-56.
14. Adams WG, Deaver KA, Cochi SL, et al. Decline of childhood Haemophilus influenzae
type b (Hib) disease in the Hib vaccine era. JAMA 1993;269:221-6.
15. Wald ER, Levine MM. Haemophilus influenzae
type b pneumonia. Arch Dis Child 1978;53:316-8.
16. Hassan-King M, Adegbola R, Baldeh I, et al. A polymerase chain reaction for the diagnosis of Haemophilus influenzae
type b disease in children and its evaluation during a vaccine trial. Pediatr Infect Dis J 1998;17:309-12.
17. Lankinen KS, Salo P, Rapola S, Salo E, Takala AK, Leinonen M. Pneumococcal capsular antigen detection after enrichment culture: an alternative to culture methods in epidemiologic research. Am J Trop Med Hyg 1997;56:211-5.
18. Salo P, Ortqvist A, Leinonen M. Diagnosis of bacteremic pneumococcal pneumonia by amplification of pneumolysin gene fragment in serum. J Infect Dis 1995;171:479-82.
19. Nohynek H, Eskola J, Kleemola M, Jalonen E, Saikku P, Leinonen M. Bacterial antibody assays in the diagnosis of acute lower respiratory tract infection in children. Pediatr Infect Dis J 1995;14:478-84.
20. Manary MJ, Lehmann D, Michael A, et al. Antigenuria in healthy Papua New Guinean children with nasal Haemophilus influenzae
type b carriage. Ann Trop Paediatr 1993;13:385-9.
21. Clemens J, Brenner R, Rao M, Tafari N, Lowe C. Evaluating new vaccines for developing countries: efficacy or effectiveness? JAMA 1996;275:390-7.
22. Shann F. Etiology of severe pneumonia in children in developing countries. Pediatr Infect Dis J 1986;5:247-52.
23. WHO. Acute respiratory infections: a forgotten pandemic. Bull WHO 1998;76(1):101-3.