American Journal of Epidemiology Advance Access originally published online on December 22, 2006
American Journal of Epidemiology 2007 165(7):756-761; doi:10.1093/aje/kwk064
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ORIGINAL CONTRIBUTIONS |
The Impact of Birth Weight on Infectious Disease Hospitalization in Childhood
From the Department of Epidemiology Research, Statens Serum Institut, Copenhagen, Denmark
Correspondence to Anders Hviid, Department of Epidemiology Research, Statens Serum Institut, Artillerivej 5, DK-2300 Copenhagen S, Denmark (e-mail: aii{at}ssi.dk).
Received for publication June 21, 2006. Accepted for publication September 18, 2006.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Low birth weight, a result of preterm birth or intrauterine growth restriction, is a well-established indicator of survival in childhood. However, corresponding epidemiologic studies of the association between low birth weight and morbidity from infections throughout childhood are sparse. The authors evaluated the relation between birth weight and infectious diseases throughout childhood in a population-based cohort study comprising all children born in Denmark from 1977 through 2004 (n = 1.7 million). Information on birth weight, gestational age, and potential confounding variables was linked to the children in the cohort, together with information on hospitalization with infectious disease. Poisson regression yielded rate ratios of hospitalization according to birth weight. The authors found that birth weight was inversely associated with risk of infectious disease hospitalization; among children aged 0–14 years, the risk of hospitalization increased 9% for each 500-g reduction in birth weight (increase in rate ratio = 1.09, 95% confidence interval: 1.09, 1.11). The effect was found to peak in infancy and to persist until 10 years of age. It was present also in children born at term (37–41 weeks of gestation). The present study is the first to demonstrate the measurable impact of birth weight on infectious diseases throughout childhood.
birth weight; child; cohort studies; communicable diseases; Denmark; hospitalization
Abbreviations: CI, confidence interval; ICD-8, International Classification of Diseases, Eighth Revision; ICD-10, International Classification of Diseases, Tenth Revision; RR, rate ratio
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Both preterm and intrauterine growth-restricted babies have depressed immune function (1–5), and epidemiologic studies have demonstrated some of the possible clinical implications of this with increased mortality from infections in low birth weight children (6, 7). However, corresponding studies of the association between birth weight and morbidity from infections are sparse.
The present population-based study evaluated the relation between birth weight and infectious disease hospitalization prospectively in a cohort comprising all children born in Denmark from 1977 through 2004, including information on birth weight, gestational age, and possible hospitalization for each child. In the study, a distinction was made between types of infectious diseases (acute upper respiratory tract infection, viral and bacterial pneumonia, septicemia, viral central nervous system infection, bacterial meningitis, and diarrhea), and the possible effects of birth weight were considered at different stages in childhood. Information on gestational age was included to determine whether a potential birth weight effect was an independent effect or a consequence of preterm birth.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Since April 1968, people living in Denmark have been given a unique identification number in the Danish Civil Registration System (8). On the basis of this registry, we constructed a cohort comprising all children born in Denmark in the period from January 1, 1977, to December 31, 2004. Using the unique personal identification number, we were able to link information on birth weight and gestational age, potential confounding variables, and information on possible hospitalization with infectious disease to children in the cohort.
Infectious disease hospitalization
Information on hospitalization with infectious disease (date of admission and diagnosis code) in the period from January 1, 1977, to December 31, 2004, was obtained from the Danish National Hospital Register (9). From 1977 to 1993, the International Classification of Diseases, Eighth Revision (ICD-8), was used to code diagnosis, and since 1994, the International Classification of Diseases, Tenth Revision (ICD-10), has been used. We included the following categories of infectious diseases in our study: acute upper respiratory infection (ICD-8 codes 460.0–466.9; ICD-10 codes J00.0–J06.9), viral pneumonia (ICD-8 codes 480.0–480.9; ICD-10 codes J12.0–J12.9), bacterial pneumonia (ICD-8 codes 481.0–482.9; ICD-10 codes J13.0–J15.9), septicemia (ICD-8 codes 038.0–038.9, 036.10; ICD-10 codes A40.0–A41.9, A39.2), viral central nervous system infections (ICD-8 codes 040.0–046.9; ICD-10 codes A80.0–A89.9), bacterial meningitis (ICD-8 codes 036.09, 320.00, 320.09, 320.19, 320.80; ICD-10 codes A39.0, G00.0–G01.9), and diarrhea (ICD-8 codes 009.0–009.9; ICD-10 codes A09.0–A09.9).
Birth weight and gestational age
Information on weight and gestational age at birth was obtained from the Danish Medical Birth Registry (10): child's birth weight (<1,000, 1,000–1,499, 1,500–1,999, 2,000–2,499, 2,500–2,999, 3,000–3,499, 3,500–3,999, 4,000–4,499, 4,500–4,999, 
5,000 g) and child's gestational age (<31, 31–33, 34–36, 37–39, 40–41, 42 weeks). The percentages of missing values for the variables birth weight and gestational age were 5.5 percent and 11.7 percent, respectively.
Potential confounders
We included information on the following possible demographic confounders in our study: sex; birth order (1, 2, 3, 4); age of mother at birth of child (<20, 20–24, 25–29, 30–34, 35 years); the degree of urbanization of the community of the household at birth (small towns/rural areas (population: <10,000), medium-sized towns (population: 10,000–99,999), large cities (population: 100,000), capital suburbs (Copenhagen suburbs), capital (Copenhagen)); and maternal nationality (Danish or not). All variables were either obtained directly from or constructed on the basis of data from the Danish Civil Registration System. From the Danish Medical Birth Registry, we further obtained information on possible confounders: 5-minute Apgar score (<6, 6–7, 8–9, 10) and cesarean section at delivery (yes, no). The percentages of missing values for the variables Apgar score and cesarean section were 8.7 percent and 9.4 percent, respectively. Information on maternal smoking while pregnant (yes, no) was available on all births from 1994.
Statistical analysis
Children in our cohort contributed person-time to follow-up from birth or January 1, 1977, whichever occurred last, until infectious disease hospitalization (using date of admission) with any of the infectious diseases listed and categorized above, death, disappearance or emigration, 15 years of age, or December 31, 2004, whichever occurred first. The resulting incidence rates according to age, calendar period, demographic factors, and birth characteristics were further analyzed with Poisson regression (log-linear regression on the infectious disease counts using the logarithm to the follow-up time as offset) to produce estimates of rate ratios (11) according to birth weight. We furthermore estimated the increase in rate ratio per 500-g reduction in weight at birth (birth weight trend). More specifically, this was done by replacing categorical index values with mean values from the category and considering the variable to be continuous.
All rate ratios were adjusted for age (0 months, 1 month, 2–5 months, 6–11 months, 1–4 years of age by 1-year intervals, 5–9 years of age, 10–14 years of age); calendar period (1-year intervals); gestational age; 5-minute Apgar score; cesarean section; sex; birth order; mother's age; degree of urbanization; and maternal nationality. 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.
All analyses were conducted by use of SAS, version 9.1, statistical software (SAS Institute, Inc., Cary, North Carolina).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
A total of 1,735,456 children were included in our cohort. The follow-up of 36,275 children was prematurely terminated because of death (n = 13,264), emigration (n = 22,488), or disappearance (n = 523). Furthermore, 3,121 children with a relevant diagnosis of infectious disease were excluded from follow-up because the date of admission was identical to the date of birth, that is, primarily nosocomial infections. During 16,259,882 person-years of follow-up, we identified 212,336 cases of infectious disease hospitalizations: 138,570 cases of acute upper respiratory tract infection, 11,352 cases of viral pneumonia, 6,912 cases of bacterial pneumonia, 4,291 cases of septicemia, 1,850 cases of central nervous system viral infections, 3,791 cases of bacterial meningitis, and 45,570 cases of diarrhea.
In table 1, we present rate ratios for categories of infectious disease hospitalization according to birth weight, together with trends. We found an increasing risk of infectious disease with reducing birth weight. Among very low birth weight children (<1,000 g), the rate ratios ranged from 1.38 (95 percent confidence interval (CI): 0.57, 3.35) for bacterial meningitis to 3.25 (95 percent CI: 2.05, 5.16) for septicemia compared with children weighing 3,000–3,499 g. For more moderate low birth weight children (e.g., 2,000–2,499 g), the rate ratios ranged from 1.26 (95 percent CI: 1.20, 1.33) for diarrhea to 1.71 (95 percent CI: 1.46, 2.00) for septicemia compared with children weighing 3,000–3,499 g. We found significant birth weight trends ranging from 1.07 (95 percent CI: 1.02, 1.12) for central nervous system viral infections to 1.16 (95 percent CI: 1.14, 1.18) for viral pneumonia, corresponding to 7 percent and 16 percent increases in the risk of infectious disease per 500-g reduction in birth weight, respectively.
|
To evaluate the impact of very low birth weight and heavy children on the trend estimates, we reestimated the birth weight trends excluding children with a birth weight of less than 1,500 g or more than 5,000 g. This changed the trend estimates to 1.07 (95 percent CI: 1.03, 1.11) for septicemia and to 1.08 (95 percent CI: 1.02, 1.13) for central nervous system viral infections; all others were unchanged.
In figure 1, we present the birth weight trend for all infections (increase in rate ratio (RR) = 1.09, 95 percent CI: 1.09, 1.11) according to age of child. We observed a significant birth weight trend from earliest infancy (increase in RR = 1.06, 95 percent CI: 1.04, 1.09, in the first month of life) until 10 years of age (increase in RR = 1.02, 95 percent CI: 1.01, 1.03, for children 5–9 years of age). The birth weight trend was greatest from 2 until 6 months of age (increase in RR = 1.21, 95 percent CI: 1.19, 1.22).
|
In figure 2, we present birth weight trends according to age of child and category of infectious disease. The pattern was similar for all categories, with a peak at 1–5 months of age and a declining but persisting effect until 10 years of age. In the earliest infancy, the only significant trend was for septicemia (increase in RR = 1.16, 95 percent CI: 1.10, 1.21). The greatest trend was for central nervous system viral infections (increase in RR = 1.39, 95 percent CI: 1.17, 1.66) in the second month of infancy. The most persistent effect was for diarrhea (increase in RR = 1.05, 95 percent CI: 1.02, 1.07, among children 5–9 years of age).
|
We reestimated the birth weight trend for all infections among children born at 37–41 weeks of gestation and found no difference compared with the trend among all children (increase in RR = 1.09, 95 percent CI: 1.08, 1.09). Similarly, we reestimated the trend for all infections among children born with 37–41 weeks of gestation according to age of child. This changed the trend estimates to 0.98 (95 percent CI: 0.95, 1.01) for children 0 months of age, 1.09 (95 percent CI: 1.06, 1.12) for children 1 month of age, 1.16 (95 percent CI: 1.14, 1.18) for children 2–5 months of age, 1.03 (95 percent CI: 1.00, 1.05) for children 4 years of age, and 1.02 (95 percent CI: 0.99, 1.05) for children 10–14 years of age; all others were unchanged.
We evaluated the potential confounding effect of maternal smoking in a subcohort with this information available, that is, children born from 1994 through 2004. We estimated the trends for all infections with and without adjustment for maternal smoking. The birth weight trend went from 1.07 (95 percent CI: 1.07, 1.08) to 1.06 (95 percent CI: 1.05, 1.07) when maternal smoking was included.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
In summary, we found inverse associations between birth weight and risk of infectious disease hospitalization throughout childhood. The risk was greatest in infancy and persisted until 10 years of age. We observed similar patterns independent of type of infectious disease. Very low or high birth weights had negligible influence on our results, and the birth weight effect was also present among children born with 37–41 weeks of gestation and, thus, was independent of prematurity.
Few epidemiologic studies of the relation between birth weight and infectious disease mortality exist. Read et al. (6) found that moderate low birth weight (1,500–2,499 g) compared with birth weights at or above 2,500 g increased the risk of infectious disease mortality by approximately 150 percent in a US cohort of 50,000 children. The effect remained in childhood and appeared to be a result of preterm birth rather than intrauterine growth restriction. Similarly, Samuelsen et al. (7) found that low birth weight (
2,499 g) compared with birth weights above that increased the risk of infectious disease mortality by approximately 70 percent in a Norwegian cohort of 1.3 million children, an effect lasting into childhood.
An adverse effect in earliest infancy of low birth weight on immunity through prematurity is well established. Preterm infants have decreased levels of immunoglobulin G compared with term infants (1). Immunoglobulin G antibodies are transferred from mother to fetus beginning at about 17 weeks of gestation, with cord-blood levels being similar to maternal titers at about 33 weeks and up to two times higher at term (3). Preterm infants have also decreased levels of complement (5).
Few studies have examined the independent effect of intrauterine growth restriction on immunity. Chatrath et al. (4) found that the lymphocyte count and percentage, CD4 percentage, immunoglobulin G concentration, and complement concentration were decreased and that the CD8 percentage was increased in small-for-gestational-age neonates compared with appropriate-for-gestational-age neonates. The authors speculated that the depressed T-cell status could be the result of thymic dysfunction, for example, thymolymphatic atrophy, which has been observed in small-for-gestational-age newborns; the high levels of CD8 a result of high levels of cortisol; the lower levels of immunoglobulin G a result of the shrunken placenta's interference with transplacental passage; and lower levels of complement a result of poor liver function. The result of this study was compatible with that of a previous older study (2).
Aaby et al. (12) found thymus size at birth to be associated with mortality from infectious diseases in infancy in Guinea-Bissau, a small country in West Africa. An interpretation of these results could be that thymus size has an influence on immunocompetence in infancy. Furthermore, since birth weight and thymus size were also associated, an association also found in Danish infants (13), low birth weight could be associated with infectious disease in infancy through reduced thymus size.
While the association between birth weight and infectious disease in infancy is biologically plausible as a result of depressed immune status, there is no readily available explanation for the persistence of the association into childhood.
Moore et al. (14) found an association between birth weight and response to vaccination in adulthood in Pakistan. Specifically, a polysaccharide typhoid vaccine was positively associated with birth weight, while rabies vaccine was not associated with birth weight. The authors suggested that antibody generation to polysaccharide antigens was compromised by fetal growth restriction, indicating that components of the immune system could be permanently programmed by events in early life.
More generally, there is support for a relation between intrauterine growth restriction and failure to thrive and stunted growth into adolescence (15). The majority of children with fetal growth restriction demonstrate catch-up growth in early childhood. However, failure-to-thrive and growth problems in a minority of children without catch-up growth could be a contributing explanation for the inverse association between birth weight and infectious disease into childhood, but this can be confirmed only in a study with specific information on postnatal growth and other factors necessary to identify failure to thrive.
A challenge in any study of birth outcomes and infectious diseases is the potential for bias and confounding. The nature of our chosen study methodology, a nationwide cohort study with historically prospective follow-up, has effectively minimized any concern over selection and information bias, as well as recall bias in particular. However, the comparability of children with respect to factors influencing the risk of infection other than the possible effect of birth weight merits discussion. We included adjustment for a number of factors (other than age and calendar period) typically associated with the risk of infection: sex, degree of urbanization, mother's nationality, mother's age at birth, and birth order. Furthermore, we included information on gestational age; 5-minute Apgar score, a strong predictor of survival; and cesarean section, an indicator of a complicated pregnancy/birth. A limitation of our study is the lack of direct socioeconomic measures and, thus, the potential for unmeasured confounding. It is likely that families of lower socioeconomic class would have both an increased risk of having lower birth weight children and infectious disease or hospitalization, potentially resulting in a false inverse association in the worst case. Although this is conceivable, the Danish society is egalitarian with free health care for everybody and less variability in the socioeconomic status of the population than, for example, the United States, limiting the magnitude and impact of potential socioeconomic confounding. Furthermore, we have already included variables that are associated with socioeconomic class, such as the mother's nationality, mother's age at birth, degree of urbanization of the place of living, and birth order (essentially the number of older siblings). Another unmeasured possible confounder in our study is maternal smoking. Babies born to mothers who smoke tobacco have lower birth weight than do babies born to mothers who do not smoke, as well as a higher risk of infection. The potential confounding effect of maternal smoking during pregnancy was evaluated in a subcohort, and we found the effect on our results to be negligible. Consequently, we believe that residual or unmeasured confounding is unlikely to explain our results.
A potential nonbiologic contributing explanation of our results could be that low birth weight children have a lower threshold for hospitalization, given the same infectious disease as normal birth weight children. However, we are unable to evaluate the contribution of this in the current study design.
Today, in developed countries, childhood mortality is low, with infectious diseases accounting for only a small percentage. Morbidity from infectious diseases, however, remains significant. The present study is the first to evaluate the impact of birth weight on infectious disease morbidity throughout childhood on this scale. Our results demonstrate the measurable impact into childhood, and they add important new information on the consequences of birth weight.
|
ACKNOWLEDGMENTS |
|---|
The study was supported by a grant from the Danish Medical Research Council.
Conflict of interest: none declared.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
- Evans HE, Akpata SO, Glass L. (1971) Serum immunoglobulin levels in premature and full-term infants. Am J Clin Pathol 56:416–18.[Web of Science][Medline]
- Ferguson AC. (1978) Prolonged impairment of cellular immunity in children with intrauterine growth retardation. J Pediatr 93:52–6.[Web of Science][Medline]
- Malek A, Sager R, Kuhn P, et al. (1996) Evolution of maternofetal transport of immunoglobulins during human pregnancy. Am J Reprod Immunol 36:248–55.
- Chatrath R, Saili A, Jain M, et al. (1997) Immune status of full-term small-for-gestational age neonates in India. J Trop Pediatr 43:345–8.
- Notarangelo LD, Chirico G, Chiara A, et al. (1984) Activity of classical and alternative pathways of complement in preterm and small for gestational age infants. Pediatr Res 18:281–5.[Web of Science][Medline]
- Read JS, Clemens JD, Klebanoff MA. (1994) Moderate low birth weight and infectious disease mortality during infancy and childhood. Am J Epidemiol 140:721–33.
[Abstract/Free Full Text] - Samuelsen SO, Magnus P, Bakketeig LS. (1998) Birth weight and mortality in childhood in Norway. Am J Epidemiol 148:983–91.
[Abstract/Free Full Text] - Malig C. (1996) The Civil Registration System in Denmark. (International Institute for Vital Registration and Statistics, Bethesda, MD) (IIVRS technical paper no. 66).
- Andersen TF, Madsen M, Jorgensen J, et al. (1999) The Danish National Hospital Register. A valuable source of data for modern health sciences. Dan Med Bull 46:263–8.[Web of Science][Medline]
- Knudsen LB and Olsen J. (1998) The Danish Medical Birth Registry. Dan Med Bull 45:320–3.[Web of Science][Medline]
- Clayton D and Hills M. (1993) Poisson and logistic regression. Statistical models in epidemiology. (Oxford University Press, Oxford, United Kingdom)227–36.
- Aaby P, Marx C, Trautner S, et al. (2002) Thymus size at birth is associated with infant mortality: a community study from Guinea-Bissau. Acta Paediatr 91:698–703.[CrossRef][Web of Science][Medline]
- Hasselbalch H, Jeppesen DL, Ersboll AK, et al. (1997) Sonographic measurement of thymic size in healthy neonates. Relation to clinical variables. Acta Radiol 38:95–8.[Web of Science][Medline]
- Moore SE, Jalil F, Ashraf R, et al. (2004) Birth weight predicts response to vaccination in adults born in an urban slum in Lahore, Pakistan. Am J Clin Nutr 80:453–9.
[Abstract/Free Full Text] - Yanney M and Marlow N. (2004) Paediatric consequences of fetal growth restriction. Semin Fetal Neonatal Med 9:411–18.[CrossRef][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. C. Wang, S. H. Lee, M. C. Lee, and L. Wang The effects of age and aboriginality on the incidence of low birth weight in mountain townships of Taiwan J Public Health, September 1, 2009; 31(3): 406 - 412. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. Tzoulaki, M.-R. Jarvelin, A.-L. Hartikainen, M. Leinonen, A. Pouta, M. Paldanius, A. Ruokonen, D. Canoy, U. Sovio, P. Saikku, et al. Size at birth, weight gain over the life course, and low-grade inflammation in young adulthood: northern Finland 1966 birth cohort study Eur. Heart J., April 9, 2008; (2008) ehn105v1. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||