Efficacy of Haemophilus influenzae type b polysaccharide-diphtheria toxoid conjugate vaccine in US children aged 18-59 months

Saturday

17

August

1991

No 8764

ORIGINAL AR TICLES

Efficacy of Haemophilus influenzae type b toxoid conjugate vaccine polysaccharide-diphtheria in US children aged 18-59 months

THE

HAEMOPHILUS INFLUENZAE VACCINE EFFICACY STUDY GROUP

prepared from the polyribosylribitol phosphate (PRP) capsule of Haemophilus influenzae b (Hib) have not consistently shown good efficacy in protecting children aged over 18

Vaccines

months from invasive Hib disease. To evaluate the efficacy of conjugate-PRP vaccines in this agegroup, and to compare their effect with that of PRP vaccines, a post-marketing case-control study was conducted among 10 400 000 persons. Between Oct 1,1988, and Feb 28, 1990, 75 patients with Hib disease and 161 control children between 18 and 60 months of age were enrolled. To minimise potentially confounding socioeconomic variables, controls were selected either from among patients’ classmates at their day-care centre or from among family acquaintances. 9 of the 75 patients had received the diphtheria toxoid conjugate Hib vaccine more than 2 weeks before onset of illness. After adjusting for age and household crowding, the efficacy of PRP Hib vaccine was 64% (95% Cl = -20, 89) and efficacy of the diphtheria toxoid conjugate Hib vaccine was 74% (95% Cl = 30, 90). The study shows that the protective efficacy of this conjugate vaccine is less than ideal and highlights the need for additional post-licensing studies to confirm and expand understanding of the efficacy of these new products. Introduction

Haemophilus inj7uenzae rype b (Hib) is the commonest of bacterial meningitis in children in the United States1 and western Europe.2 11 000 to 15 000 cases of

cause

invasive Hib disease occur each year in the United States in

children

aged under 5 years. 3-6% of these die and long-term neurological sequelae of meningitis have been documented in up to 30% of survivors of meningitis. Antibody to the polyribosylribitol phosphate capsule (PRP) protects against invasive disease. However, in man PRP functions as a T-cell independent antigen, so the immunological response to vaccines composed of PRP alone is highly age-dependent. A PRP vaccine was 90% effective in children 18 months of age and older in a large prospective trial in Finlandbut several studies done after widespread use of PRP vaccines in US children aged 18-59 months showed only moderate efficacy (ranging from - 55% to 80%).’ Conjugation of specific proteins to PRP greatly improves immunogenicity in children 18-59 months of age,8-lO and in the past 3 years, three conjugate vaccines have been licensed in children of this age in the United StatesHaemophilus influenzae type b polysaccharide-diphtheria toxoid conjugate vaccine (PRP-D, Connaught Laboratories, Inc, ’Prohibit’); the CRMl97 protein conjugate (HbOC, Praxis Biologics, ’Hibtiter’), the protein being a non-toxic variant of the diphtheria toxin molecule isolated from a mutant Corynebacterium diphtheriae strain; and the meningococcal outer membrane protein complex conjugate (PRP-OMP, Merck, Sharp and Dohme, ’Pedvaxhib’). Here we report our post-marketing casecontrol study done to evaluate the efficacy of PRP and PRP-protein conjugate vaccines in use between October, 1988, and February, 1990. for

use

ADDRESSES: Meningitis and Special Pathogens Branch (J. D. Wenger, MD, R Pierce, MPH, K. A. Deaver, C. V. Broome, MD), Biostastistics and Information Management Branch (B. D. Plikaytis, MS), and Respiratory Diseases Branch (R. R. Facklam, PhD), Division of Bacterial and Mycotic Diseases, National Center for Infectious Disease, Centers for Disease Control, Atlanta, Georgia 30333, USA. Correspondence to: Dr. J. D. Wenger, Meningitis and Special Pathogens Branch, National Centersfor Disease Control, Building 1, Room 4413, C09 Atlanta, GA 30333, USA.

396

Methods Surveillance for invasive Hib disease

TABLE I-CHARACTERISTICS OF CHILDREN WITH CASES OF INVASIVE HIB DISEASE BY ENROLMENT STATUS

Laboratory-based surveillance for

all cases of invasive Hib conducted in four areas-the 8 contiguous counties of metropolitan Atlanta, Georgia; the state of Oklahoma; 4 metropolitan counties in the state of Tennessee; and 3 contiguous counties in the San Francisco Bay area, California. From intercensal estimates the total population of the surveillance area was 10 400 000 in mid-1989. Surveillance coordinators in each area identified someone in each acute-care hospital in their area to inform them of cases of invasive Hib disease-illness in which Hib was isolated from a normally sterile site (such as cerebrospinal fluid or blood). H influenzae isolates were serotyped at the local laboratory, and most isolates were also sent to the state or Centers for Disease Control (CDC) laboratories for confirmation. For each case a form containing demographic, medical, and bacteriological data was sent to the surveillance coordinator, who reviewed the data and entered the information into a computerised database. Every month the data were sent to CDC, where they were further edited and analysed. The sensitivity of the surveillance system was evaluated through audit of laboratory logs in a sample of hospitals in each area, by visiting the hospital or by reviewing their computerised records. beds and 17% of remaining 51% of all hospitals with >200 hospitals were audited. The sensitivity of the surveillance system for Hib disease was calculated using the following formula: ([number of cases detected by active surveillance in audited hospitals]/[number of cases detected by active surveillance and laboratory audit in audited hospitals]) x 100.

disease

was

Patient and control selection Parents or guardians of all children who had invasive Hib disease between ages 18-5 and 60 months were approached by telephone, mail, or in person to participate in the study. Up to 3 control subjects were enrolled for each participant. Various factors related to socioeconomic status-such as household crowding,l1 black race,12 and number of elementaryschool-aged children13-influence risk of Hib disease. Although it is not clear exactly what socioeconomic variables are risk factors in specific populations, their presence plus the effect of socioeconomic status on likelihood of being vaccinated can confound estimates of vaccine efficacy in a case-control analysis. For example, children with low socioeconomic status are generally overrepresented among cases. If controls are then randomly selected from the general population, they are more likely to be vaccinated because of the overall higher socioeconomic status and better medical care, and the resulting estimate of efficacy will be falsely high. If the confounding risk factor is unknown (or not determined in cases and controls). adjustments are impossible. To reduce the chance of such confounding, we enrolled controls from among children likely to have similar socioeconomic backgrounds as the patients. For children in day care-ie, being cared for with at least one other unrelated child for 4 or more hours in the week before hospital admission for Hib disease-which is a well-documented risk factor for diseases controls were enrolled from among their classmates. A roster of all children in the classroom on the day the patient fell ill was prepared, ordered by age of child, and parents’ or guardians’ approval was sought in order of closeness in age of the classmates to the patient, until up to 3 controls were enrolled for each patient. For 6 day-care cases poor compliance by day-care centres resulted in control selection which began with children not necessarily closest in age to the case. If the patient did not attend day-care, the parents/guardians were asked for a list of acquaintances with children aged between 185 and 60 months, and controls were sought in the same manner as for day-care classroom control children.

information on Hib vaccination (ie, brand of vaccine, lot numbers, date of vaccination), other immunisations, and the child’s underlying medical condition. To verify telephone information, photocopies of charts were obtained for children whose Hib disease occurred after vaccination with an Hib conjugate vaccine.

Data

analysis

Vaccine efficacy was calculated with the following formula: vaccine efficacy = ( 1 - odds ratio for vaccination) x 100. Efficacy could range from 100% to a negative number. For example, if children who received the vaccine did not acquire the disease, whereas unvaccinated children did, the point estimate of the vaccine would be 100%. If the rate of disease among vaccinated children was exactly the same as among unvaccinated children, the point estimate of efficacy would be 0%. If vaccinated children were more likely than unvaccinated children to acquire the disease, the point estimate of efficacy would be a negative number. The matched nature of the data was maintained in the analysis of vaccine efficacy by use of conditional multiple logistic regression. Since patients and controls received several different vaccines, each of which has some protective effect, efficacy calculations for individual vaccines were made by using subjects who had received no other vaccine as the comparison group. Since age and crowding were associated with disease in univariate analyses, the final model used to determine efficacy of individual vaccines included these variables. The statistical significance of differences between proportions was calculated by use of the X2 statistic (or Fisher’s exact test where

appropriate). Results 422 episodes of invasive Hib disease were identified in children less than 5 years of age between Oct 1, 1988, and Feb 28, 1990, in the surveillance areas. 95 of these were in the study age-range (18-5 to 59 months, inclusive). The sensitivity of the surveillance system for disease in the study age-range was 97%. Completed sets of case and control information were obtained for 75 of the 95 cases (79%). Reasons for failure to enrol were inability to contact the patient’s parents/guardians (6 cases), refusal of parents/ guardians or day care centre director to participate (6 cases), and failure to identify appropriate controls (8 cases). Patients enrolled in the study (n = 75) did not differ significantly from patients who were not enrolled (table 1). Data were obtained on a mean of 2controls for each case. All eligible controls (up to 3 closest in age to the patient) were enrolled for 50 cases (67%%); all but 1 eligible control for 14 (19%); and all eligible controls but 2 for 5 (7%). None of TABLE N—RiSK FACTORS FOR INVASIVE HIB DISEASE IN ENROLLED CASES AND CONTROLS

Data collection

Questionnaires were administered to parents/guardians by telephone, to obtain information on socioeconomic status, family structure, and medical care. The medical-care-givers (eg, private physicians, public clinics) who had at any time given the child a vaccination were also administered a telephone questionnaire for

HS high school

397

TABLE III-VACCINATION STATUS OF ENROLLED CASES AND CONTROLS

(95 % CI 14,93) and for non-meningitic Hib disease it was 51% (95% CI = -160, 91). Efficacy of PR1"-D for all invasive Hib disease was 67% (95% CI = - 5, 90) in day-care-related cases and 85 % (95 % CI -18, 98) in non-day-care cases. =

=

The number of cases occurring in any one surveillance area

precise estimates of the efficacy by geographic area were not possible. The univariate estimate of PRP-D efficacy in metropolitan Atlanta was 100% (lower 95% CI = - 27); in Oklahoma, 100% (lower 95 % CI =61%), in the San Francisco Bay area 66 % (95 % CI = - 75, 94); and in 900, 74). Tennessee, - 68% (95% CI We reanalysed the data, making several adjustments to identify the effect of specific biases on our estimate of vaccine efficacy. 13% of controls were not among the 3 determined to be closest in age to the patient. When these controls were dropped from the analysis, efficacy of PRP-D in the multivariate model was 70% (95% CI = 16, 89). 7 children in the study age-group had H influenzae disease but was

the first three

eligible controls were enrolled for 6 cases (8%). Altogether, 87% of controls were among the 3 potential controls closest in age to the patient. Failure to enrol potential controls was due to inability to locate control parents/guardians or their refusal to participate. Although controls were matched on the basis of closeness in age to the patients, their difference in mean age, though small, was significant (table 11). Crowding (number of persons in a houshold divided by the number of rooms in the house) was also significantly associated with disease. Vaccination status among patients and controls is shown in table III. A child was considered vaccinated if he/she received vaccine more than 14 days before date of admission of the case for that particular case-control set. Because no difference has been shown between the three PRP vaccines marketed widely in the United States in terms of efficacy or immunogenicity, they were evaluated as a single PRP vaccine ("PRP, any" in table III). However, the conjugate vaccines were treated separately because they have defmite structural and immunological differences. No study subjects received PRP-OMP during the study period. Table IV shows characteristics of the 9 cases, which occurred in children vaccinated with PRP-D. 1 of these children died from Hib disease. None of these children had any previously recognized immunological disorders or underlying diseases except for 1 child who had Down’s syndrome. 8 of the 9 children with vaccine failure had been vaccinated before October, 1988, compared with 17 of 42 controls (p<002). One explanation for this difference be variation in vaccine efficacy by production date or might lot number. Although no vaccine lot or group of lots was significantly associated with disease in vaccine recipients, 3 of 8 vaccinated patients (38%), compared with 4 of 38 vaccinated controls (11%), received vaccine prepared from a specific bulk lot (p 0-09). The final multivariate model for vaccine efficacy included age, crowding, and vaccination status with a specific vaccine or group of vaccines (ie, PRP vaccines). The point estimate 20, of efficacy for the PRP vaccines was 64% (95% CI 89). The point estimate of PRP-D efficacy was 74% (95% =

=

-

CI = 30, 90). In a multivariate model including crowding and age in months, the efficacy of PRP-D for Hib meningitis was 76% TABLE IV-CHARACTERISTICS OF CHILDREN ACQUIRING DISEASE AFTER IMMUNISATION WITH PRP-D

SF=San Francsco,TN= Tennessee, OK—Oklahoma

small,

so

=

-

isolates were lost and information on serotype could not be obtained. Attempts to enrol these children resulted in 5 additional case-control sets. With these children in the analysis, the multivariate model showed an efficacy of 75%

(95%CI=33,91). Discussion

efficacy of PRP-D in children 18-59 months of age higher than that of PRP vaccines in the same population, which is consistent with immunogenicity studies showing superior immunogenicity of the conjugate vaccine compared with the PRP vaccine. However, the point estimate of efficacy for PRP-D was lower than that obtained in several other post-marketing case-control The

was

studies. 15-18 Our control selection differed from that employed in the other studies. Greenberg et al,15 using controls identified by random digit dialing, found an efficacy of 88 % (95 % CI 42,97). Osterholm16 and Shapiro and Wald18 used age-matched controls identified from birth records and found efficacies of 96 % (95 % CI = 65,99) and 84% (95% Cm = 28, 97), respectively. Although it is impossible to tell whether the difference in point estimates is due to control selection, since not more than one method was used in one population, the direction of the difference is consistent with our hypothesis that the net effect of systematic selection of controls from the general population =

digit dialing or birth date matching) overestimate of the vaccine efficacy. For instance, in this study, crowding (defined as > 1 person/ room) was significantly associated with disease, and controls from crowded households were significantly less likely to have received PRP-D (13% vs 30% vaccinated, p<005). We could control for this effect since it was evaluated in the study, but if there are other similar but unknown (or undetermined, if information on these factors was not requested from study participants) risk factors, no adjustments in the analysis could be made. One reason why our estimate may be lower than others’ is that we approached these potential confounding variables by matching patients with controls from their own socioeconomic milieu. Additional studies should be done with different methods of control selection to establish the validity of each method. Such studies are important since two other conjugate vaccines (PRP-OMP and HbOC) have recently been licensed for use in infants 2 months of age and older in the United States. 19.20 Determination of the age-specific efficacy (whether by would be

random

an

398

and duration of

protection of each of these products will require additional post-licensing case-control studies. Accurate interpretations of the results of these studies will be critical for future recommendations for vaccine use. The immunogenicity of PRP-D vaccine in children 18-59 months of age is significantly better than that of the PRP vaccine. However, even in this age-group, levels of antibody after a single dose are age dependent. One study in United States children showed a significant difference by age in geometric mean antibody titres of children given a single dose of PRP-D vaccine between 15 and 24 months of age. In addition, in only 58% to 78% of these children was a level of 1 0 Ilg!ml of antibody to PRP reached; this level is thought to be indicative of long-term protection. The efficacy demonstrated in the current study is consistent with such immunogenicity data. 62% (26 of 42) of controls who received PRP-D vaccine were immunised at 24 months of age or less, as were 5 of the 9 children with vaccine failure. The efficacy of PRP-D in cases and controls aged 24 months or less at time of disease was 44% (95% CI - -260, 91), compared with 77% (95% CI - 19, 94) for older children, although this difference was not statistically significant. The influence of other factors on vaccine failure was investigated. A significantly higher proportion of children who were vaccinated before a specific date had vaccine failure. One possible explanation for this is variability in efficacy due to differences between vaccine lots. Persons receiving a particular bulk lot did have a higher proportion of failure than those receiving other lots, although the difference was not statistically significant (p=009). Pustlicensing surveillance may be helpful in identifying variation in vaccine lot efficacy. The influence of host factors in vaccine failures in our study is also being evaluated. No child had an underlying disease specifically associated with Hib disease. One child was an American Indian from Oklahoma and thus may have been at some increased risk of disease, but this case did not affect the point estimate of vaccine efficacy since the single matched control had also received PRP-D. 4 of the 9 children with vaccine failure were from Tennessee. There was no temporal or spatial relation between these cases to suggest transmission of a particularly virulent strain. Additional analysis of the strains and comparisons with invasive disease strains from the other areas are now under way. Since the end of the (Feb 28, 1990), 2 additional cases of vaccine failure have been identified in the surveillance areas,1 in metropolitan Atlanta (the only area without any vaccine failures during the study period) and another in Tennessee. We identified no "early failures" or cases of Hib disease occurring less than 2 weeks after immunisation with PRP-D vaccine. This observation is consistent with other studies, which suggests that immunisation with Hib conjugate vaccines is not associated with an increased risk of invasive Hib disease, as was questioned with PRP immunisations.21 We could not evaluate risk of early disease after PRP vaccination since by the time our study began physicians had discontinued use of this product. The level of efficacy demonstrated in this study (protection of 74% of vaccinated children 18-59 months of age) is not ideal. More effective vaccine preparations for children of this age may be available, and the results of this study emphasise the need for continued evaluation of different conjugate vaccines.

study

Haerriophilus mj1Uatzae Vavcinc Efficacy Study Group: Dr Arrhur Remgold (University of California at Berkeley, Berkeley), Ms Grelchen

Anderson (San Francisco Department of Health, San Francisco) and Ms Elizabeth Stone (Alameda County Deparunent of Health, Oakland, California); Dr Lewis Lefkowitz, Dr Mane R. Griffin, Ms Jo A. Taylor, Ms Margarel S. Rados (Department of Preventive Medicine, Vanderbilt University, Nashville, ’1 autessee); Dr David S. Stephens, Dr MOllica M. Farley, Mr Chriswphei Harvey, and Dr ’I’ina M. Stull, MD (Emory University Department of Medicine, Atlanta, Georgia); Dr Gregory R. Istre, Dr Scott J. N. McNabb, Ms Pam Archer, and Ms Jane Strack (Oklahoma State Department of Health, Oklahoma City, Oklahoma); Ms Nan Pigott, Ms Gail Bosley, Dr John A. Elliot, Ms Ruth Franklin, and Mr Ray L. Ransom (Centers for Disease Control, Atlanta, Georgia).

REFERENCES 1.

Wenger JD, Hightower AW, Broome CV, and the Bacterial Meningitis Study Group. Bacterial meningitis in the United States: report of a multistate surveillance study. J Infect Dis 1990; 162: 1316-323.

2. Noah ND. Epidemiology of bacterial meningitis: UK and USA. In: Bacterial meningitis. London: Academic Press, 1987: 93-115. 3. Peltola H, Kayhty H, Virtanen M, Makela PH. Prevention of Haemophilus influenzae type b bacteremic infection with the capsular polysaccharide vaccine. N Engl J Med 1984; 310: 1561-66. 4. Black SB, Shinefeld HR, Hiatt RA, Fireman BH, Kaiser Permanente Pediatric Vaccine Study Group. Efficacy of Haemophilus influenzae type b capsular polysaccharide vaccine. Pediatr Infect Dis 1988; 7: 149-56. 5. Shapiro ED, Murphy TV, Wald ER, Brady CA. The protective efficacy of Haemophilus b polysaccharide vaccine. JAMA 1988; 260: 1419-22. 6. Harrison LH, Broome CV, Hightower AW, et al. A day-care-based study of the efficacy of Haemophilus b polysaccharide vaccine. JAMA 1988; 260: 1413-18. 7. Osterholm MT, Rambeck JH, White KE, Makela PH. Lack of efficacy of Haemophilus b polysaccharide vaccine in Minnesota. JAMA 1988; 260: 1423-28. 8. Berkowitz CD, Ward JI, Meier K, et al. Safety and immunogenicity of Haemophilus influenzae type b polysaccharide and polysaccharide diphtheria toxoid conjugate vaccines in children 15 to 24 months of age. J Pediatr 1987; 110: 509-14. 9. Madore DV, Johnson CL, Phipps DC, et al. Safety and immunogenicity of Haemophilus influenzae type b oligosaccharide-CRM197 conjugate vaccine in infants aged 15 to 23 months. Pediatrics 1990;; 86: 527-34. 10. Shapiro ED, Capobianco LA, Berg AT, Zitt MQ. The immunogenicity of Haemophilus influenzae type b polysaccharide-Neisseria meningitidis group B outer membrane protein complex vaccine in infants and young children. J Infect Dis 1989; 160: 1064-67. 11. Cochi SL, Fleming DW, Hightower AW, et al. Primary invasive Haemophilus influenzae type b disease. a population-based assessment of risk factors. JPediatr 1986; 108: 887-96. 12. Fraser DW, Mitchell JE, Silverman LP, Feldman RA. Risk factors in bacterial meningitis: Charleston County, South Carolina. J Infect Dis

1973; 127: 271-77. 13. Istre GR, Conner JS, Broome CV, et al. Risk factors for primary invasive Haemophilus influenzae disease: increased risk from day care attendance and school-aged household members. J Pediatr 1985; 106: 190-95. 14. Redmond SR, Pichichero ME. Haemophilus influenzae type b disease, an epidemiologic study with special reference to day-care centers. JAMA 1984; 252: 2581-84. 15. Greenberg DP, Vadheim CM, Bordenave N, et al. Protective efficacy of Haemophilus influenzae type b polysaccharide and conjugate vaccines in children 18 months of age and older. JAMA 1991; 265: 987-91. 16. Osterholm MT. Efficacy of Haemophilus b plain polysaccharide (PRP) vaccine and conjugate vaccine (PRP-D) in Minnesota. Abstracts of the 29th Interscience Conference on Antimicrobial Agents and Chemotherapy. September 18, 1989; Houston, Texas. Abstract 449A. 17. Murphy TV, Herrin-Kane CM, Coury S, Medley F. Protective efficacy of Haemophilus influenzae type b vaccines in Dallas County, Texas. Clin Res 1990; 38: 183A. 18. Shapiro ED, Wald ER. The protective efficacy of PRP-D conjugate vaccine against Haemophilus influenzae type b (Hib). Abstracts of the 30th Interscience Conference on Antimicrobial Agents and Chemotherapy. October 23, 1990; Atlanta, Georgia. Abstract 604. 19. Immunization Practices Advisory Committee. Haemophilus b conjugate vaccines for prevention of Haemophilus influenzae type b disease among infants and children two months of age and older. MMWR 1991; 40 (no. RR-1): 1-7. 20. Black SB, Shinefeld HR, Fireman B, et al. 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. 21. Ward JI, Broome CV, Harrison LH, Shinefeld HR, Black SB. Haemophilus influenzae type b vaccine: lessens for the future. Pediatrics 1988, 81: 886-93.