Abstract
Objectives. To estimate the association between birth weight and hearing impairment among Norwegians born between 1967 and 1993, taking other pregnancy-related conditions into consideration.
Methods. A cohort study was conducted of all Norwegian live births from 1967 to 1993 (n = 1 548 429) linking information of the Medical Birth Registry of Norway and the register for the National Insurance Administration, which covers all Norwegians. The Medical Birth Registry of Norway has recorded information on birth weight and other pregnancy-related conditions as well as diseases of the mother before and during pregnancy. The register of the National Insurance Administration contains information on all Norwegians who have received cash benefits for a disease/disability, including hearing impairment. Data up to 1997 are included; thus, the follow-up period varies between 29 and 3 years.
Results. The occurrence of hearing impairment was 11 per 10 000, decreasing from 60 per 10 000 for birth weights <1500 g to 6 per 10 000 for birth weights >4499g. Compared with birth weights between 3000 g and 3499 g, the adjusted rate ratio of hearing impairment was 7.55 (95% confidence interval: 4.81–11.87) for birth weights <1500 g and 0.50 (95% confidence interval: 0.34–0.73) for birth weights >4499 g. The association did not change substantially with adjustment for other pregnancy-related conditions. Restricting the analyses to term born, the association between hearing impairment and low birth weight became stronger.
Conclusions. Birth weight was a strong predictor of hearing impairment in the Norwegian population. Children who were born at term with a low birth weight seemed to be a particularly vulnerable group.
Hearing impairment in children is a relatively common and serious problem.1–12 People who have severe hearing impairment will in general improve their condition with hearing aids, cochlear implants, and/or special training. Depending on the diagnostic criteria, the studied age group, and calculation method, the occurrence of hearing impairment varies,10 but a commonly reported figure is 10 to 20 per 10 000 young people.1–12 Several studies have shown that children who at birth had very low birth weight have increased risk of severe hearing impairment.13–15 Less attention has been given the association between birth weight and hearing impairment in general taking into consideration other pregnancy-related conditions, which may represent causal factors in the development of hearing. A large and representative population is therefore required to assess the effects of birth weight per se. The present study is based on information on pregnancy outcomes and other pregnancy-related conditions from the Medical Birth Registry of Norway (MBRN) and includes all live-born children in Norway between 1967 and 1993. The information was linked with the National Insurance Administration Register (NIAR), which covers the total population. The register includes information on all people who have received cash benefits to cover expenses related to treatment and habilitation of hearing impairment. The aim of this study was to estimate the association between birth weight and hearing impairment among Norwegians born between 1967 and 1993, taking into consideration other pregnancy-related conditions.
METHODS
Study Population and Data Sources
The MBRN comprises all live and still births in Norway after 1966.16 Children who were born alive between 1967 and 1993 were included in the study (n = 1 548 429). MBRN was linked with NIAR. The NIAR contains information on all residents of Norway who have received cash benefits from the National Insurance Administration.17 Information was updated through 1996, which means a follow-up period between 29 and 3 years. All Norwegians are granted cash benefits if they have a disease/disability that gives them substantial treatment expenses, including expenses to treat hearing impairment. Insurance membership and income taxes are paid together. The unique national identification number made the linkage feasible. For registration of diagnoses, The Norwegian Register of Disability used International Classification of Diseases, Seventh Revision (ICD-7) codes through 1982 and ICD-9 codes for the rest of the period, whereas the MBRN used ICD-8 codes throughout the period. The National Bureau of Statistics provided data for the highest level of education among Norwegian women who gave birth between 1967 and 1993. Information on education had been updated in 1970, in 1980, and on a yearly basis from 1985 to 1993. Education level registered nearest in time after the birth of the child was used in the analyses. The national identification number linked the information to the other information. Norwegians who were born between 1967 and 1993 were mainly of white origin (>95%).18
Variables
Hearing Impairment
A person who was registered by the NIAR with an ICD-7 code of 398 before 1983 or an ICD-9 of 389 after 1982 was defined as having hearing impairment. Only people who can document that they have substantial expenses for treatment of their disease/disorder will receive cash benefits and be registered in NIAR. All institutions that diagnose and treat hearing impairment or adapt and sell hearing aids are aware of the rights of these patients. They will inform them about their rights and may even apply for economic support on their behalf, ensuring that these people get state benefits.
Pregnancy Outcomes
Birth weight was registered in grams and grouped into 8 strata (<1500 g, 1500–1999 g, 2000–2499 g, 2500–2999 g, 3000–3499 g, 3500–3999 g, 4000–4499 g, and 4500 g+); gestational age was registered in completed weeks and grouped into 6 strata (<28 weeks, 28–33 weeks, 34–36 weeks, 37–39 weeks, 40–42 weeks, and >42 weeks). Plurality was registered as singleton births or multiple births. Congenital malformations (ICD-8: 740–759) were also registered.
Other Pregnancy-Related Conditions
Maternal age (<21 years, 21–30 years, 31–35 years, and 36+ years), parity (1, 2, 3, 4, 5+), maternal education (<11 years, 11–14 years, 15 years or more), year of birth (1967–1975, 1976–1984, 1985–1993), single-parent family at birth, year and months of first receiving cash benefits, and year and month of emigration or death were recorded.
Diseases and Complications During Pregnancy
The following conditions were considered in the analyses: preeclampsia (including eclampsia, toxemia; ICD-8: 637), hyperemesis (ICD-8: 638), other pregnancy complication (ICD-8: 630–634), and maternal viral infections with exanthema during pregnancy (ICD-8: 050–057, including varicellae-, herpes simplex-, morbilli-, and rubella infections).
Information on maternal diseases and pregnancy complications were dichotomized. Missing values for birth weight and birth weights <500 g were excluded from the analyses (n = 2571) together with missing values for parity (n = 5582) and maternal age (n = 7; in total 7900). Gestational age <16 weeks and >44 weeks was recorded as missing values. Birth weight of 3000 g to 3499 g and gestational age 40 to 42 weeks were used as reference categories for these 2 variables, ensuring large reference categories including mainly people born from normal pregnancies. Year of birth stratified as above was included as a covariate in all of the adjusted analyses.
Statistical Methods
Cox regression was used to estimate crude rate ratios of hearing impairment in univariate analyses and adjusted rate ratios (aRRs) adjusted for confounders and to assess possible interactions. Observation time was ended when a person had received cash benefits, when a person had died or emigrated, or at the end of 1996. The proportional hazard assumption of Cox regression was tested by running separate regressions for people who had receiving cash benefits for hearing impairment before and after the age of 5 years. Because of a high number of missing values for gestational age (n = 112 304) and maternal education (n = 47 870), missing was coded as a separate layer and kept in the analyses. Otherwise, people with missing values were excluded. People who had died or emigrated before the age of 5 months were excluded (n = 14 000), as their chance of having been evaluated for and had a diagnosis of hearing impairment was very low. This does not affect the estimated rate ratios because Cox regression uses data only from subjects who are at risk at date of receiving cash benefits and no cases were defined before 5 months of age. Data analyses without people with observational time <3 years gave similar results as the main approach and are not presented. To look for cohort effects, we performed stratified analyses for people who were born in 3 equal time periods (before 1976, 1976–1984, and after 1984). Attributable fraction indicating the fraction of hearing impairment that could have been prevented if birth weights were elevated above 2499 g and above 3499 g was estimated.19
RESULTS
Figure 1 shows hearing impairment per 10 000 Norwegians by year of birth 1967 to 1993 in total and for those who had receiving cash benefit for hearing impairment before and after 5 years of age. In total, 11 per 10 000 had received cash benefits, and 5 per 10 000 had received cash benefits before the age of 5 years. Hearing impairment varied by year of birth with the lowest occurrence in the youngest cohorts. This tendency was not apparent in those who had receiving support before the age of 5 years and is probably caused by shorter follow-up in the youngest cohorts. The median age for first-time cash benefits for hearing impairment was 5 years, 718 (45%) received cash benefits from before they were 5 years of age, 633 (39%) between 5 and 9 years, 208 (13%) between 10 and 14 years, and 52 (3%) when they were 15 years of age or older.
Hearing impairment per 10 000 among Norwegians born between 1967 and 1993 and followed to 1996 by year of birth in total and for people who had received cash benefits for hearing impairment before the age of 5 years (<5 years) and later (>4 years).
Among Norwegians who were born with a birth weight <1500 g, 60 per 10 000 were registered as having a hearing impairment (Table 1). The corresponding figure for children who were born with a birth weight of 4500 g or more was 6 per 10 000 people. There was a consistent reduction in the occurrence of hearing impairment by increasing birth weight. A similar effect of birth weight was seen when the outcome was restricted to children who had receiving cash benefit for hearing impairment before and after the age of 5 years (figures not shown). Table 1 also gives the occurrence of hearing impairment per 10 000 for other pregnancy outcomes and pregnancy-related conditions. Virus infection with exanthema during pregnancy increased the risk of hearing impairment in the offspring (aRR: 14.86; 95% confidence interval [CI]: 11.42–19.32). The rate ratios for hearing impairment by birth weight were mainly unchanged in a model adjusting for other pregnancy outcomes and other pregnancy-related conditions. Contrary to this, the increased risks related to short gestational age were reduced when adjusted for the other conditions. Missing information on gestational age was associated with a slightly increased risk of hearing impairment. Mean birth weight for this group was 3486 g (standard deviation: 600), whereas it was 3496 g (standard deviation: 577) in the total material. Being born with a malformation increased the risk of hearing impairment both in crude and adjusted analyses (crude rate ratio: 2.73 [95% CI: 1.20–3.38]; aRR: 2.64 [95% CI: 2.13–2.28]), whereas different types of pregnancy complications did not show any statistically significant association with hearing impairment. Hearing impairment was slightly more common among boys than girls and in multiple births compared with single births. However, when controlling for birth weight, multiple births did not increase the risk of hearing impairment. Birth order and maternal age were also associated with the risk of hearing impairment.
Occurrence of Hearing Impairment (per 10 000) and the cRR and aRR With 95% CI for Hearing Impairment by Birth Weight, Gestational Age, and Other Pregnancy-Related Conditions in Norwegians Born Between 1967 and 1993 and Followed up to 1996
Table 2 shows similar trends for the association between birth weight and hearing impairment for 1967 to 1975, 1976 to 1984, and 1985 to 1993 even if the risk became somewhat weaker in the youngest cohorts. It is interesting that the aRR for hearing impairment by virus infection with exanthema during pregnancy dropped from 41.55 (95% CI: 30.30–56.97) for the period 1967 to 1975, to 8.41 (95% CI: 4.62–15.28) for 1976 to 1984, to 3.04 (95% CI: 1.26–7.34) for 1985 to 1993. These infections were registered for 0.18%, 0.25%, and 0.35% of the pregnancies in the corresponding periods.
aRR* With 95% CI for Hearing Impairment by Birth Weight Among Norwegian-Born Children 1967 to 1975, 1976 to 1984, and 1985 to 1993 and Followed up to 1996
Table 3 shows that the association between birth weight and hearing impairment was not reduced when only children who were born with a gestational age of 37 weeks or more were kept in the analysis. For preterm children, the association was less evident. Because of relatively few cases in the extreme groups, the 2 smallest and the 2 largest weight categories were collapsed for the presentation in Fig 2. Figure 2 also indicates that the effect of low birth weight was stronger among full-term than preterm infants. There is also a tendency that preterm heavy infants had a higher risk of hearing impairment compared with heavy infants born at term.
Hearing impairment by birth weight in preterm and term-born Norwegians who were born between 1967 and 1993 and followed to 1996.
Occurrence of Hearing Impairment and the aRR* With 95% CI for Hearing Impairment by Birth Weight in Preterm and Full-Term Norwegian-Born Children 1967 to 1993
The population attributable fraction of hearing impairment when everyone with birth weight <2500 g was moved to the 2500 g to 2999 g weight group was 8%, and 25% when everyone with birth weight <3500 g was moved to the 3500 g to 3999 g weight group.
DISCUSSION
The risk of developing hearing impairment increased consistently with decreasing birth weight. The trend was strongest for the lowest and leveled off for the highest birth weights. The effect was robust to adjustment for a number of possible confounders, including gestational age, virus infection with exanthema during pregnancy, and malformations. It was particularly evident among infants who were born at term.
Birth weight has been shown to be associated with the risk of hearing impairment,13–15 but few studies have explored the relation in more detail for the whole specter of birth weights, taking into account other pregnancy-related conditions. We estimated the association between birth weight and hearing impairment in the total Norwegian population born 1967 to 1993 including >1 500 000 people. It is probably the largest population-based study of its kind. The short observation time for the youngest cohorts will result in a slight underestimation of the true occurrence of hearing impairment. However, the observed occurrence of hearing impairment is well in accordance with other studies.1–12 The outcome was based on information on people who had received cash benefits from National Insurance Administration to cover expenses in connection with necessary treatment and/or training for hearing impairment. As explained earlier, it seems reasonable to consider this as a valid measure for severe hearing impairment among younger Norwegians. A few people received cash benefits for hearing impairment for the first time when they were 15 years of age or older. The reason for this could have been late onset or late diagnosis of hearing impairment. There was no way to decide which explanation was correct. However, the few cases of late onset could not have affected the association between birth weights and hearing impairment. The association between hearing impairment and birth weight in people who had received cash benefits before the age of 5 years indicated the same association with birth weight as for the total material.
An association between the risk of hearing impairment and birth weight was also evident for birth weights considered as normal. On the basis of these findings, one might speculate that reduced fetal growth, as a result of genetic disposition or environmental exposures, is important for the development of hearing. There was no apparent inverse J-shaped curve for the association between birth weight and hearing impairment as seen for birth weight and perinatal mortality.20 However, low statistical power as a result of a small number of children with high birth weight, even in this large sample, reduces the chance to draw a firm conclusion. Malformations and virus infections with exanthema during pregnancy were as expected associated with impaired hearing. However, these factors did not affect the association between birth weight and impaired hearing. The interesting reduced effect on hearing impairment during the observation time of virus infections with exanthema illustrates that causes of hearing impairment may change. The change took place even when reported cases of infections increased during the study period. The main explanation for this is probably the introduction of vaccination against rubella during the study period. However, increased reports of nonrubella and less severe infections with exanthema could also have weakened the association.
Improvement in peri- and postnatal care during the last decades might have resulted in increased survival among children at increased risk of developing adverse health effects such as hearing impairment. It might also have resulted in better care and treatment and reduced risk of hearing impairment of the same group of newborns. Results from analyses of the different cohorts could not clearly suggest which of these effects was the strongest as the associations between hearing impairment and birth weight were only slightly reduced among the youngest children.
In crude analyses, low gestational age bore a higher risk of developing hearing impairment, but this was no longer evident when controlled for birth weight. One could speculate that the results suggest intrauterine growth retardation as a risk factor. However, to include both birth weight and gestational age in the same regression model is questionable as low birth weight and short gestational age might indicate the same growth disturbances. For other health outcomes, the residual distribution of birth weights seems to be important for the risk of adverse health outcomes.20 Children of low gestational age are the main source of the residuals on the left side of the distribution. In this case, children who are born at term with low birth weight, indicating intrauterine growth retardation, seemed to be a more vulnerable group than preterm children with low birth weight. It was also a weak tendency to the opposite relation for normal and high birth weights. In some way, growth in utero in total or high growth velocity seems to be favorable for the development of hearing. Whether growth, as measured in this study, indicates genetic predisposition or environmental exposures affecting the risk of hearing impairment is impossible to say. Missing information on gestational age slightly increased the risk of having hearing impairment. The effect did not seem to have anything to do with birth weight, and we had no information that made it possible to explore this finding further.
Low birth weight is clearly a risk factor for hearing impairment and should, as often is the case, warrant special attention. However, the estimated attributable fractions showed that the lowest birth weight groups contributed only moderately to the total number of hearing impairment. These measures also show that it is difficult to define a meaningful birth weight limit for a screening program for newborns. This is further complicated by the fact that the birth weight distribution varies between populations and that a certain low birth weight in one population might not lead to the same increase in adverse health effects in different populations.
Several other pregnancy-related conditions were associated with the risk of hearing impairment in some of the analyses, including gender, multiple pregnancies, maternal age, and birth order. Detailed analysis of this relation was not the topic of this study, as these variables did not substantially affect the association between birth weight and hearing impairment.
This study clearly demonstrated that low birth weight is associated with increased risk of developing hearing impairment. The effect was particularly strong and evident among full-term children. The results support the view that reduced fetal growth in total and/or reduced growth velocity increases the risk of developing hearing impairment. One could speculate that slow weight gain increases the time interval that a fetus is vulnerable to certain harmful exposures. Even if small children have a substantial increased risk of hearing impairment, screening for hearing impairment among newborns below a certain birth weight will be of limited value as a large proportion of children with hearing impairment are born with normal birth weight.
REFERENCES
- Copyright © 2002 by the American Academy of Pediatrics