Childhood Vaccination and Nontargeted Infectious Disease Hospitalization FREE

During the last 2 decades, more vaccinations have become available and routine vaccination schedules have become increasingly complex. This has led to concern among some that multiple antigen vaccines, such as the measles-mumps-rubella vaccine, or aggregated vaccine exposure could lead to immune dysfunction, resulting in infectious diseases not targeted by vaccination, occurring as a result of an “overload” mechanism. In a 2002 safety review of multiple immunizations and immune dysfunction, the US Institute of Medicine concluded that there was strong evidence for the existence of biological mechanisms by which multiple vaccinations could influence the risk of nontargeted infectious diseases.¹ However, epidemiological and clinical support for the effect was lacking, and some studies even favored a beneficial effect on nontargeted infectious diseases.^2- 6

To test the hypotheses that multiple-antigen vaccines or aggregated vaccine exposure increase risk of nontargeted infectious diseases, we evaluated the relationship between routinely administered childhood vaccines and nontargeted infectious diseases in a large population-based cohort study comprising all children born in Denmark from 1990 through 2001. Our study included longitudinal information on type and number of vaccine doses received and hospitalization with infectious diseases, specifically acute upper respiratory tract infection, viral and bacterial pneumonia, septicemia, viral central nervous system infections, bacterial meningitis, and diarrhea. Included vaccines were Haemophilus influenzae type b (Hib); diphtheria-tetanus-inactivated poliovirus; diphtheria-tetanus-acellular pertussis-inactivated poliovirus; whole-cell pertussis; measles-mumps-rubella (MMR); and oral poliovirus vaccine–the majority of these being multiple-antigen vaccines.

Since April 1968, people living in Denmark have been given a unique identification number in the Danish Civil Registration System.⁷ This registry keeps daily updated information regarding demographic features on all residents including dates of birth, death, emigration, and change of address. From this registry we constructed a cohort of all children born in Denmark from January 1, 1990, through December 31, 2001. Using the unique personal identification number, we were able to link information on all childhood vaccinations, infectious disease hospitalization, and potential confounders to the children in the cohort. Since our study was registry-based with no active participation from study subjects, no approval from an ethics committee was required according to Danish law. The use of registry data on individual subjects was approved by the Danish Data Protection Agency.

During the study period of 1990 through 2001, the Danish childhood vaccination program included vaccinations against pertussis, measles, mumps, rubella, diphtheria, tetanus, polio, and Haemophilus influenza type b. A detailed overview of the childhood vaccines used in the Danish program in this period to 5 years of age is given in Table 1. The dates of vaccination with the first, second, or third dose of vaccines were obtained from the National Board of Health, as previously described.^8- 9 In Denmark, only general practitioners administer childhood vaccinations and they are reimbursed when reporting these to the National Board of Health. The National Board of Health has kept a registry of these reports since 1990. Data on MMR has only been available since September 1991, and thus, children born in 1990 were classified as having unknown MMR status.

Information on hospitalization with infectious diseases in the period January 1, 1990, to December 31, 2001, was obtained from the Danish National Hospital Register.¹⁰ During 1990 to 1993, ICD-8 (International Classification of Diseases, 8th Revision) was used, and during 1994 to 2001, ICD-10 (International Classification of Diseases, 10th Revision). We included the following categories of infectious diseases in our study: acute upper respiratory tract infection (ICD-8: 460.xx, 461.xx, 462.xx, 463.xx, 464.xx, 465.xx, 466.xx; ICD-10: J00.x, J01.x, J02.x, J03.x, J04.x, J05.x, J06.x); viral pneumonia (ICD-8: 480.xx; ICD-10: J12.x); bacterial pneumonia (ICD-8: 481.xx, 482.xx; ICD-10: J13.x, J14.x, J15.x); septicemia (ICD-8: 038.xx, 036.10; ICD-10: A40.x, A41.x, A39.2); viral central nervous system infections (ICD-8: 040.xx, 041.xx, 042.xx, 043.xx, 044.xx, 045.xx, 046.xx; ICD-10: A8x.x); bacterial meningitis (ICD-8: 036.09, 320.00, 320.09, 320.19, 320.80; ICD-10: A39.0, G00.x, G01.x); and diarrhea (ICD-8: 009.xx; ICD-10: A09.x).

Children in the cohort contributed person-time to follow-up from birth until hospitalization with any of the infectious diseases, death, disappearance or emigration, 5 years of age, or December 31, 2001, whichever occurred first. The resulting incidence rates for infectious disease hospitalization were analyzed with Poisson regression (log-linear regression on the incidences using the logarithm to the follow-up time as offset), producing estimates of incidence rate ratios (RRs) according to vaccination status.¹¹ Vaccination status was considered a time-varying variable, ie, children could contribute person-time as both unvaccinated and vaccinated. The time following receipt of a vaccine was subdivided into 3 periods: within 14 days, 14 days after until 3 months after, and more than 3 months after. To eliminate confounding resulting from vaccination being postponed for acutely ill children, the within-14-days period was considered a lag period; follow-up time and case activity from this period was not attributed to either the vaccinated or unvaccinated group in any analyses unless stated explicitly. We estimated the dose-response relation between the vaccines and infectious disease hospitalization as the increase in RR per dose (per 0.5 mL in the case of the whole-cell pertussis vaccine) and only among vaccinated children. Thus, we estimated the possible effect of vaccination on infectious disease hospitalization through: (1) RRs comparing vaccinated with unvaccinated children; (2) increases in RRs per vaccine dose among vaccinated children; (3) RRs comparing children in the period 14 days after to 3 months after any vaccine dose with unvaccinated children; and (4) RRs comparing children in the period more than 3 months after any vaccine dose with unvaccinated children. These RRs were always adjusted for age (<1 year of age, 1-month intervals; 1-2 years of age, 3-month intervals; 2-4 years of age, 1 year intervals), calendar period (1-year intervals), and receipt of other vaccines when estimating the previously stated RRs. Furthermore, to take into account a possible change in the distribution of age at infectious disease hospitalization from ICD-8 to ICD-10, we included an age and calendar period interaction (0 months of age, 1-5 months of age, 6-11 months of age, 1 year of age, and 2-4 years of age for the periods 1990-1993 and 1994-2001). All analyses were conducted using SAS statistical software version 8.2 (SAS Institute Inc, Cary, NC).

Information on possible confounding factors was obtained from the Danish Civil Registration System, the Danish Medical Birth Registry,¹² and the National Hospital Register: child’s sex, child’s place of birth (Copenhagen, Copenhagen suburbs, area with at least 100 000 population, area with population of 10 000-99 999, area with population <10 000), child’s birth weight (<2500 g, 2500-2999 g, 3000-3499 g, 3500-3999 g, ≥4000 g), mother’s country of birth (Denmark or other), mother’s age at birth of child (<20 years, 20-24 years, 25-29 years, 30-34 years, 35-39 years, ≥40 years), and birth order (1, 2, 3, ≥4). Month of follow-up was also considered as a potential confounding variable. The percentage of missing values for the variables birth weight, child’s place of birth, and mother’s country of birth were 4.06%, 0.03%, and 0.82% respectively. Variables were identified as confounders for the association between vaccination and infectious disease hospitalization and included in the regression models if they changed the RR comparing vaccinated with unvaccinated children by more than 10% for a given vaccine. When adjusting for the potential confounding effect of variables with missing values, we used the method of single imputation, replacing a missing value with the most common value.

A total of 805 206 children were included in the cohort. During 2 900 463 person-years of follow-up, 84 317 cases of infectious disease hospitalization were identified. The follow-up of 12 552 children was prematurely terminated because of death (n = 4681), emigration (n = 7710), or disappearance (n = 161).

Table 2 presents the person-years of follow-up time and number of cases of infectious disease hospitalization according to age, calendar period, sex, and vaccination status.

Figure 1 displays the RRs for infectious disease hospitalizations comparing vaccinated with unvaccinated children and increases in RRs for infectious disease per dose of vaccine among vaccinated children. The confounding effects of sex, place of birth, birth weight, mother’s country of birth, mother’s age at birth, birth order, and season were negligible. Figure 2 presents RRs comparing vaccinated children in the period from 14 days to 3 months after vaccination with unvaccinated children, and RRs comparing vaccinated children in the period more than 3 months after vaccination with unvaccinated children.¹¹ The only statistically significant adverse association between any combination of vaccine type and infectious disease hospitalization was between the Hib vaccine and acute upper respiratory tract infection (RR comparing vaccinated with unvaccinated children, 1.05; 95% confidence interval [CI], 1.01-1.08; RR comparing children >3 months after any vaccine dose with unvaccinated children, 1.07; 95% CI, 1.03-1.12 with no dose-response association). Thus, of the 42 possible associations between the 6 vaccines and the 7 nontargeted infectious disease outcome categories, only 1 indicated an adverse effect. At the 5% level of statistical significance, 1 adverse association of 40 possible would be expected by chance alone.

Of the RRs estimated for children in the within-14-days lag period relative to unvaccinated children, the only adverse association was between MMR vaccine and acute upper respiratory tract infection (RR, 1.10; 95% CI, 1.01-1.21). We reestimated RRs comparing vaccinated with unvaccinated children for all associations including this period in the vaccinated group. No RR comparing vaccinated with unvaccinated children was increased more than 10%.

In our main analyses, children were censored when hospitalized with any of the infectious disease categories. In a further analysis, we reestimated RRs comparing vaccinated with unvaccinated children in which other infectious disease categories than the one under analysis did not result in censoring but were instead a time-varying exposure. Including this previous infectious disease hospitalization exposure as a possible confounding variable increased only 2 RRs comparing vaccinated with unvaccinated children by 10% or more: oral poliovirus vaccine and viral central nervous system infection (RR comparing vaccinated with unvaccinated children, 0.93; 95% CI, 0.53-1.62 increased to 1.16; 95% CI, 0.68-1.97), oral poliovirus vaccine and diarrhea (RR comparing vaccinated with unvaccinated children, 0.88; 95% CI, 0.82-0.95 increased to 0.99; 95% CI, 0.72-1.34).

A minor group of the unvaccinated children consists of children whose parents have chosen not to have their children vaccinated at all. To evaluate the impact of this group, we focused on the effect of vaccination after the age of 12 months and compared RRs of vaccinated with unvaccinated children with and without excluding completely unvaccinated children. None of the RRs comparing vaccinated with unvaccinated children were increased by more than 10%.

We evaluated the association between the total number of vaccine doses received (aggregated vaccine exposure, N = 0 . . . 13) and infectious disease hospitalization as the increase in RR per dose. These dose-responses were RR, 0.99 (95% CI, 0.99-0.99) for acute upper respiratory tract infection; RR, 0.94 (95% CI, 0.93-0.95) for viral pneumonia; RR, 0.96 (95% CI, 0.95-0.97) for bacterial pneumonia; RR, 0.98 (95% CI, 0.96-1.00) for septicemia; RR, 0.99 (95% CI, 0.96-1.02) for viral central nervous system infections; RR, 1.00 (95% CI, 0.98-1.02) for bacterial meningitis; and RR, 0.99 (95% CI, 0.98-0.99) for diarrhea. Furthermore, we compared children vaccinated at least once with completely unvaccinated children. These RRs were 0.82 (95% CI, 0.78-0.86) for acute upper respiratory tract infection; RR, 0.70 (95% CI, 0.64-0.77) for viral pneumonia; RR, 0.79 (95% CI, 0.64-0.98) for bacterial pneumonia; RR, 0.53 (95% CI, 0.41-0.67) for septicemia; RR, 1.47 (95% CI, 0.90-2.40) for viral central nervous system infections; RR, 0.84 (95% CI, 0.60-1.18) for bacterial meningitis; and RR, 0.90 (95% CI, 0.83-0.99) for diarrhea.

We evaluated the relationship between routinely administered childhood vaccines and nontargeted infectious disease hospitalizations in a large population-based cohort study. We specifically tested 2 hypotheses: whether multiple-antigen vaccines increase the risk of nontargeted infectious diseases, and whether aggregated vaccine exposure increases the risk of nontargeted infectious diseases. Of the associations between specific vaccines and nontargeted infectious disease hospitalizations, we found an adverse association between the Hib vaccine and acute upper respiratory tract infections. This association, if causal, has limited clinical relevance because of the modest magnitude of the effect. It also has no bearing on the specific hypothesis tested due to Hib being a single-antigen vaccine. However, the association is unlikely to be causal because it did not present as either a temporal effect or a dose-response effect. Furthermore, this 1 adverse association of a possible 42 is within the limits of what would be expected purely by chance in a study with multiple comparisons at this level of statistical significance. Conversely, the 15 observed protective associations suggest that vaccination may have nontargeted protective effects. When considering aggregated vaccine exposure, we found no adverse associations between an increasing number of vaccinations and nontargeted infectious disease hospitalizations. Overall, our results support neither of the tested hypotheses. However, using only hospitalization, presumably representing the more severe cases of infectious disease, as our study outcome is a limitation. When evaluating the generalizability of our results, the differences in schedule between Denmark and other countries (vaccinations are administered later in Denmark) should be taken into consideration.

A major challenge in any study of vaccination and nontargeted infectious diseases is the potential for bias and confounding. Our study methodology, a nationwide cohort study with prospective follow-up, has effectively minimized any concern over selection and information bias, particularly recall bias. However, the comparability of vaccinated and unvaccinated children with respect to factors influencing risk of infection other than the possible effect of vaccination merits discussion. We included a number of factors (other than age, calendar period, and receipt of other vaccines) typically associated with risk of infection: sex, place of birth, birth weight, mother’s country of birth, mother’s age at birth, birth order, and month of the year. The confounding effect of these factors was negligible. In other studies, few factors have been identified in Denmark as being associated with obtaining vaccinations.¹³ A series of programmatic changes (introduction of new vaccines and old vaccine types replacing new ones) in the Danish childhood vaccination program during the study period also allayed concern about comparability of the vaccinated and unvaccinated groups. For the majority of the vaccines in the study, the unvaccinated group was primarily composed of children vaccinated with other vaccines. Thus, for a given vaccine, we compared 2 groups of vaccinated children differing only on receipt of that vaccine. This eliminates concern over the comparability of an unvaccinated group of children whose parents have chosen not to have their child vaccinated at all. Furthermore, identifying and eliminating this small group of children (defined by us as children still completely unvaccinated after the age of 12 months) had minimal impact on our results. Finally, transient confounding by indication could potentially result in protective effects of vaccination in the period after vaccination if vaccination of acutely ill children was postponed. Consequently we introduced a 14-day lag period after vaccination and adjusted for prior infectious disease hospitalization. Neither of these measures changed our results, indicating little if any confounding by indication affecting the overall results.

Few studies specifically examining the association between vaccination and nontargeted infectious diseases exist. A number of US studies from the early 1990s have examined the association between diphtheria, tetanus, whole-cell pertussis vaccine and invasive bacterial disease.^2- 5 None of these studies reported an increased risk of nontargeted infectious disease after vaccination. However, the studies were small case-control studies^2- 4 and a cohort study without an unvaccinated comparison group.⁵ Furthermore, the majority of these studies only included the immediate postvaccination period. Otto and colleagues conducted a randomized controlled trial comparing vaccinated and unvaccinated children with respect to nonspecific morbidity, such as coughing or signs of rhinitis, and found that nonspecific morbidity was significantly more common among unvaccinated children.⁶ Despite being a randomized controlled trial, this study had some methodological shortcomings; the study was not blinded and nonspecific morbidity was self-reported through a parent diary. A recent study by Miller and colleagues examined the association between MMR vaccination and bacterial infection and found a protective effect using the self-controlled case series method.¹⁴

Outside the United States and Europe, a study from a small West African country examined the association between vaccination (BCG; polio; diphtheria, tetanus, whole-cell pertussis; and measles) and childhood mortality and found that diphtheria, tetanus, whole-cell pertussis vaccine was associated with increased childhood mortality.¹⁵ Recent studies commissioned by the World Health Organization¹⁶ have not confirmed this effect.^17- 19

In conclusion, our results do not support the hypothesis of increased risk of nontargeted infectious disease hospitalization after childhood vaccination.