OBJECTIVE: The goal of this study was to investigate the association of poor birth condition with long-term social and economic outcomes at 25 to 31 years of age.
METHODS: This was a population-based cohort study using data derived from linkage of routinely collected Swedish data. All term infants born in Sweden between 1973 and 1979 identified from the Swedish birth registry (n = 651 615) were included in the study. Infants were categorized into 3 groups: (1) infants with a normal (>7) Apgar score at 1 or 5 minutes of age without encephalopathy; (2) infants with a low (<7) Apgar score at 1 and 5 minutes of age without encephalopathy; and (3) infants with a low (<7) Apgar score at 1 and 5 minutes with evidence of encephalopathy. The main outcome measures were achievement of a university education and participant's income in early adulthood.
RESULTS: Infants with low Apgar scores who did not develop encephalopathy were less likely to have attended university (odds ratio [OR]: 1.14 [95% confidence interval (CI): 1.05–1.23]) and were more likely to have no income from work (OR: 1.19 [95% CI: 1.07–1.32]) than those born in good condition. Infants who developed encephalopathy also had greater risks of these adverse outcomes (not attended university, OR: 1.94 [95% CI: 1.13–3.33]); no income from work, OR: 3.08 [95% CI: 1.89–5.01]).
CONCLUSIONS: Infants born in poor condition had worse measures of social performance than their peers, and this association was not restricted to those infants who developed obvious neurologic symptoms in the neonatal period. However, even in infants with likely encephalopathy, more than half obtained employment and one third attended university.
WHAT'S KNOWN ON THIS SUBJECT:
Severe perinatal asphyxia is known to have devastating effects on the newborn infant. The long-term outcome of perinatal asphyxia on social and economic outcomes is unknown.
WHAT THIS STUDY ADDS:
Although the effect sizes were small, these study infants with brief perinatal compromise had measurably worse social outcomes. Infants who developed encephalopathy showed far greater impact, yet many of these infants attended university and gained employment in adulthood.
Perinatal asphyxia can have profound and devastating effects on the newborn infant; however, its long-term impact on social and economic outcomes in adulthood is unknown. Although it is well recognized that infants who develop encephalopathy after a period of perinatal asphyxia have increased risk of learning difficulties, we found that even infants with only transient poor condition at birth perform worse than their peers on tests of cognition in childhood1 and early adulthood.2 However, although research of novel therapies for perinatal asphyxia is currently a priority, surprisingly little is known about its long-term social and economic consequences. In addition to its effects on IQ, there is some evidence that infants exposed to perinatal asphyxia have increased risks of psychiatric disease,3 autism,4 and behavioral and educational problems.5 Despite these reported associations, the potential impact on long-term pragmatic measures of functioning are unknown.
The a priori aim of this study was to investigate, in term infants, the effects of poor condition at birth on measures of education, employment, earnings, and socioeconomic position in adulthood in a cohort of ∼600 000 Swedish people.
The data set was based on the birth records of all term (≥37 completed weeks') infants born in Sweden between 1973 and 1979 identified from the Swedish Medical Birth Register.6
In total, the cohort consisted of 651 615 infants. The birth registry provides data on 98% to 99% of Swedish births7 and includes Apgar scores, and both neonatal and maternal diagnoses (coded using the International Classification of Diseases, Eight Revision). Although there is no agreed measure for perinatal asphyxia, poor condition at birth (as measured by the Apgar score) is a commonly used indicator of hypoxia-ischemia.8 Linkage using each Swedish citizen's unique identification number to the Longitudinal Integration Database on Social Insurance and Labor Market (recorded in 1999 and 2004) allowed access to the educational, employment, and marital status of the study subjects. Annual income is recorded in Swedish krona (1000 Sk equals ∼US$142). Further linkage to the Population and Housing Census of 1980, 1985, and 1990 provided information on parental education and socioeconomic position.
Infants from multiple births or those with cardiovascular, respiratory, neurologic, or multiple system congenital abnormalities were removed from the data set. Infants with weights, lengths, or head circumferences at birth >5 SDs from the mean (corrected for gestational age) were considered improbable entries and, together with subjects with missing data, were removed.
Analysis Strategy and Study Power
Information on birth condition and perinatal well-being was retrieved from the birth registry. Neurologic symptoms were identified using the diagnosis in the register coded according to the International Classification of Diseases, Eighth Revision. A recorded diagnosis of seizures, encephalopathy, or brain injury caused by asphyxia or with unspecified cause was considered evidence of encephalopathy.
Three groups of infants were identified: (1) infants with a normal (≥7) Apgar score at 1 or 5 minutes of age without encephalopathy (“reference” group); (2) infants with a low (<7) Apgar score at 1 and 5 minutes of age without encephalopathy (“low Apgar” group); and (3) infants with a low (<7) Apgar score at 1 and 5 minutes with evidence of encephalopathy (“encephalopathy” group). Infants with encephalopathy but normal Apgar scores were considered likely to have other pathologic conditions and were not used in the main analysis.
The 2 primary outcomes were the participants' income in early adulthood and if they obtained a university education. Other secondary outcomes were: employment status (employed versus no employment); receipt of disability pension, handicap allowance, or sickness benefit; and resident with parents and marital status (married/cohabiting, single, or living with parents).
For all outcome measures, data from 2004 (when the participants were 25–31 years old) were used where possible. If unavailable, 1999 data were used instead (∼2% of subjects, depending on outcome measure).
In addition, other risk factors were included in the analyses as potential confounders, divided into 3 groups: antenatal—gender, maternal parity, birth weight, length and head circumference (as z scores), and preeclampsia; intrapartum/neonatal—mode of delivery (cesarean delivery, instrumental [vacuum extraction or forceps], or unassisted vaginal delivery), maternal and neonatal infection, and gestational age; and social factors—maternal and paternal age (<20 years, 20–24 years, 30–34 years, and >34 years), maternal occupation, and education derived from the Population and Housing Census (if maternal details were missing, paternal measures were used if available).
Initially subjects with and without missing data were compared. The distribution of risk factors and potential confounders was investigated between the infants in the 3 exposure groups. Unadjusted associations between birth condition and social, economic, and educational measures were derived. To assess the association of birth condition with the various outcomes, logistic regression (for binary outcomes) and multinomial regression models (using the Stata mlogit command for outcomes with ≥3 categories [income quartile and marital status]) were used. Results are given as odds ratios (OR) for a poor outcome (not attending university or not earning money) compared with the reference group. Adjustment for possible confounders was performed by adding the variables described above to the models in blocks of common variables (defined above). Participant's age at the measurement of the outcomes and gender were controlled for in all models. We also conducted tests of statistical interaction to investigate whether the association between birth condition and social outcome may differ depending on the subject's gender by adding an appropriate interaction term to the model. Comparison between models was made using the likelihood ratio test. In a proportion of the male population (n = 146 712), a measure of cognition (an IQ score based on 4 subtests measuring logical, verbal, spatial, and technical abilities) was available at the age of 18 years from a conscription examination and has been presented previously.2 To investigate possible causal pathways, the analysis was repeated in this subgroup of the cohort adjusting for their cognitive skills.
Three sensitivity analyses were performed. To assess those infants who did not have evidence of poor condition at birth but who still progressed to develop encephalopathy, the analysis was repeated including infants in the encephalopathy category even if they had apparently normal Apgar scores at birth. An additional analysis was performed restricted to those infants with both full maternal and paternal socioeconomic and education data. Finally, because of missing data (∼9% of subjects), the analysis was repeated using an imputation technique (multiple imputation using chained equations) to impute missing values. Because of technical limitations only a random 10% of the reference group infants were included in this analysis. All infants with poor condition at birth were included.
All analyses were conducted with Stata 10 software (Stata Corp, College Station, TX), and results are presented as number (percent), mean (SD), or median (interquartile range) as appropriate. Ethical approval was obtained from Southmead Research Ethics Committee, Bristol, United Kingdom.
The initial data set contained data for 651 615 infants born at ≥37 weeks' gestational age. Infants from multiple births (n = 7558), with congenital abnormalities (n = 5380), and those who had evidence of encephalopathy without preceding low Apgar scores (n = 448) were removed from the data set, leaving a total of 638 229 eligible infants.
A total of 45 174 (7%) infants had missing or improbable data on 1 or more covariates, leaving a total of 593 055 infants. Because not all infants had full data on outcome measures, the proportion included in the analyses varied between 583 881 (91% of the eligible population) and 587 771 (92%). Infants excluded from at least 1 analysis (n = 56 156) either because of missing or improbable covariate (n = 45 174) or outcome (n = 25 825) data were more likely to have low Apgar scores (1.6% vs 0.5%; P < .001) and develop encephalopathy (0.10% vs 0.01%; P < .001). They were less likely to have a university education (34% vs 44%; P < .001) and had a lower median income (123 600 [6300–239 100] vs 177 500 [51 000–254 500] Sk; P < .001) in early adulthood.
The prevalence of low (<7) 1 and 5-minute Apgar scores was 3376 (0.6%). Most covariates were associated with the risk of poor birth condition (Table 1). Infants in poor condition at birth were more likely to be male, to be born by cesarean delivery, had higher rates of neonatal sepsis, and lower birth weights (all comparisons, P < .001). Mothers were more likely to be primiparous (P < .001), but there was no statistically significant difference in parents' socioeconomic profile across the 3 groups (P = .19). There was only weak evidence that parental university attendance was associated with the infant's birth condition (P = .06). The mean age of the participants at the assessment of educational achievement and income was 28 years.
In the univariable analyses, infants in poor condition at birth were less likely to be married or cohabiting, and more likely to be unemployed or claiming handicap, disability, or sickness benefits (Table 2). There was some evidence that infants with low Apgar scores (P < .001) or encephalopathy (P = .064) were less likely to have attended university. Infants with low Apgar scores seemed to have a median income similar to the reference group (176 900 vs 175 700 Sk per annum; P = .326); infants who developed encephalopathy had the lowest median income of the 3 groups (110 400 Sk per annum; P = .002) (Table 2).
Table 3 shows the results of the regression analyses. In the final model, infants with low Apgar scores who did not develop encephalopathy were less likely to have attended university (OR: 1.14 [95% confidence interval (CI):1.05–1.23]), were more likely to be in the lowest income quartile (OR: 1.11 [95% CI: 1.00–1.24]) with evidence for a trend (Plinear = .02) and to have no income from work (OR: 1.19 [95% CI: 1.07–1.32]). They were also more likely to be living with their parents (OR: 1.21 [95% CI: 1.06–1.37]) or claiming handicap, disability, or sickness benefits (OR: 1.25 [95% CI: 1.14–1.37]) than infants in good condition at birth. In the final model, there was little evidence for a difference in the risk of unemployment (P = .38). Infants with encephalopathy were less likely to have attended university (OR: 1.94 [95% CI: 1.13–3.33]) and more likely to live alone (OR: 2.22 [95% CI: 1.16–4.26]) or with their parents (OR: 4.27 [95% CI: 1.96–9.31]), more likely to be claiming handicap, disability, or sickness benefits (OR: 4.55 [95% CI: 2.85–7.25]) and more likely to be unemployed (OR: 2.70 [95% CI: 1.72–4.24]) than infants in good condition at birth. There was some evidence that infants with encephalopathy were more likely to be in the lowest income quartile (OR: 1.86 [95% CI: 0.99–3.47]) with strong evidence for a trend (Plinear = 0.009) and to have no income from work (OR: 3.08 [95% CI: 1.89–5.01]). There was little evidence that the association between birth condition and educational status (Pinteraction = .88) or income (Pinteraction =.66) differed according to gender.
Repeating the analysis in the subset of men with cognitive measures available produced compatible (although imprecise) results to the main analysis (OR of not attending university: low Apgar score, OR: 1.26 [95% CI: 1.08–1.46], P = .003; encephalopathic, OR: 1.28 [95% CI: 0.34–4.90], P = .72). Adding the cognitive scores into the models attenuated the results substantially (OR of not attending university: low Apgar score, OR: 1.16 [95% CI: 0.98–1.38], P = .08; encephalopathic, OR: 0.83 [95% CI: 0.19–3.61], P = .80). Repeating the analysis including infants in the encephalopathy category even if they had apparently normal Apgar scores at birth produced similar results to the main analysis (OR of not attending university, OR: 1.94 [95% CI: 1.54–2.44], P < .001), as did a repeat analysis using only those adults with both maternal and paternal social economic data (OR of not attending university, low Apgar score OR: 1.13 [95% CI: 1.03–1.22], P = .006; encephalopathic, OR: 1.85 [95% CI: 1.07–3.19], P = .03). The final sensitivity analysis, imputing missing data, also produced results consistent with the main (complete case) analysis (OR of not attending university: low Apgar score, OR: 1.15 [95% CI: 1.06–1.24], P < .001; encephalopathic, OR: 2.21 [95% CI: 1.41–3.49], P < .001).
We found that poor birth condition was associated with a range of adverse long-term social outcomes. This association was not restricted to those infants who developed encephalopathy but was also present in those who seemed to develop no neurologic symptoms in the neonatal period. Although outcomes were worse in infants who developed neurologic signs in the neonatal period, one third still attended university and two thirds were employed.
The association between encephalopathy in the newborn period and neurodisability is well recognized,9 although there are little published data on long-term outcomes. We have previously reported the intellectual outcomes of a Swedish cohort2 as well as a cohort from the United Kingdom1 who were born in poor condition. In both cases, infants born in poor condition, even if they did not develop neurologic signs, had lower IQ scores than a reference group of infants born in good condition, although the effects were modest. The present analysis indicates that these associations persist into adulthood, producing measurable differences in social performance. Additional data support the concept of a continuum of reproductive casualty,10 such that although profound perinatal events cause death or obvious neurologic deficit, milder insults may cause subtle defects in functioning, not evident at birth, and that are only detectable as the child grows older. Controlling for IQ measured at age 18 years (in male subjects) attenuated the associations, suggesting that some of the associations seen here may be partially mediated by the impact of poor birth condition on cognitive performance, although it is possible that IQ is acting as a marker for other (unmeasured) effects of asphyxia on brain functioning. It should be noted that recent publications suggest that the use of 100% oxygen at birth has been associated with worse outcomes than the use of room air.11 Because the infants in our study were born in an era when oxygen was used during resuscitation, it is therefore possible that a component of the adverse outcomes may be because of hyperoxic damage, secondary to the resuscitation process itself.
The present analysis was based on a very large cohort, and the proportion of infants with low Apgar scores in our population (0.6%) seems consistent with that reported elsewhere (0.8%12), although it is likely that the scoring of Apgar scores depends on a degree of observer subjectivity.13 Using a birth registry to identify infants with encephalopathy may have missed some infants with neurologic signs and hence introduced bias, although previous work with this registry has suggested it is a reliable data set.6 We also excluded infants who developed neurologic signs but who did not have low Apgar scores on the grounds that these infants may not have had a significant hypoxic-ischemic cause to their encephalopathy,14 although a sensitivity analysis including them produced results compatible with the main analysis. Our aim was to identify infants with a persistently low Apgar score to be as specific as possible for infants born in poor condition (rather than those who respond to simple resuscitative measures). In a previous analysis of linked data from the Swedish birth registry, we suggested that small cognitive deficits may exist even in infants previously thought to recover quickly from poor birth condition.2 In this supplementary work, we have taken a commonly used definition of poor condition at birth (a low Apgar score at 5 minutes) to quantify the long-term outcomes for those infants who do, or do not, develop encephalopathy.
In addition, there were few missing data, and a sensitivity analysis using a multiple imputation technique also produced similar results to the main analysis. We had detailed information on a number of potential confounders and although a degree of residual confounding is likely, we saw little change in the point estimates after adjusting for the covariates, suggesting this is unlikely to be an important limitation. It is also possible that prenatal diseases, such as metabolic defects, not controlled for in these analyses, may present with poor Apgar scores and predispose to later cognitive dysfunction. In addition, the data were routinely collected and although measurement bias is possible, it seems unlikely that any bias would be differential to the classification of birth condition used and hence is likely to only attenuate the results seen here.
The results seem consistent with 1 publication that presented data on the social outcomes of infants after developing encephalopathy. Kjellmer et al15 reported the outcomes of 53 infants who developed encephalopathy after needing resuscitation at birth and compared them with 30 controls. Infants born in poor condition were more likely to live with parents (34% vs 17%) or be unemployed (15% vs 7%), but because of small numbers, the results were correspondingly imprecise. One important (but necessary) limitation is that the present analysis is based on infants born 30 years ago, and therefore the results may have less relevance to infants born more recently, particularly for those infants receiving novel neuroprotective therapies.16
In this cohort of Swedish infants, those born in poor condition (as indexed by a low Apgar score at 5 minutes) had worse measures of social performance than those born in good condition. These infants were less likely to attend university and more likely to not be earning money, still living with parents, and be claiming handicap, disability, or sickness benefits. This association was not restricted to those infants who develop obvious neurologic symptoms in the neonatal period, further supporting the concept of a continuum of reproductive casualty,10 although effect sizes were small. The outcome of infants who developed encephalopathy was unsurprisingly even worse, although a substantial proportion of these infants seemed to attend university and gained employment in early adulthood.
- Accepted February 8, 2011.
- Address correspondence to Finn Rasmussen, PhD, Child and Adolescent Public Health Epidemiology, Department of Public Health Sciences, Karolinska Institutet, Norrbacka SE-17176, Stockholm, Sweden. E-mail:
All of the authors made substantial contributions to the conception and design, analysis, and interpretation of data; drafted the article and revised it critically for important intellectual content; and approved the final version.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- OR =
- odds ratio •
- CI =
- confidence interval
- Odd DE,
- Rasmussen F,
- Gunnell D,
- Lewis G,
- Whitelaw A
- Dalman C,
- Thomas HV,
- David AS,
- Gentz J,
- Lewis G,
- Allebeck P
- Marlow N,
- Rose AS,
- Rands CE,
- Draper ES
- Sweden S
- Copyright © 2011 by the American Academy of Pediatrics