OBJECTIVE. Increasing numbers of infants born preterm survive into adulthood. In this study, we analyzed the effect of having been born preterm on disability and vocational success in young adults.
METHODS. A Swedish national cohort of 522310 infants born in 1973–1979 were followed up for disabilities and income in national registers in 2002 at the age of 23 to 29. Hypotheses were tested in multivariate analysis with logistic regression models on the log scale for dichotomized outcomes and linear regression for continuous variables.
RESULTS. There was a stepwise increase in disability in young adulthood with increasing degree of preterm birth. A total of 13.2% of children born at 24 to 28 weeks' gestation and 5.6% born at 29 to 32 weeks' gestation received economic assistance from society because of handicap or persistent illness, which is which is equivalent to nearly 4 times the risk of those born at term after adjustment for socioeconomic and perinatal confounders. Moderate (33–36 weeks' gestation) and marginal (37–38 weeks' gestation) preterm birth also carried significantly increased risks for disability and were responsible for 74% of the total disability associated with preterm birth. Preterm birth was associated with a lower chance of completing a university education and a lower net salary in a stepwise manner. The total economic gain for Swedish society, in terms of taxes and decreased costs for benefits, if all long-term effects of preterm birth could have been prevented in the birth cohorts in this study, would have amounted to 65 million euros in 2002 alone.
CONCLUSIONS. The majority of adults who were born very preterm lived an independent and self-supportive life. Moderately preterm birth carries a considerable risk for long-term impairment. There are strong economic incentives for secondary prevention of disability associated with preterm birth.
Increasing numbers of children born preterm survive into adulthood as a consequence of progress in prenatal and neonatal care. A considerable number of cohorts of former patients from NICUs have been created and followed prospectively over time.1 At school age 10% to 12% have been described to have considerable impairment because of a neurologic disability,2,3 as well as low cognitive test scores and an increased risk of behavior disturbances including attention-deficit/hyperactivity disorder.4 A recent follow-up at 20 years of age demonstrated a slightly lower IQ at 20 years of age and a persistent educational disadvantage in preterm with a birth weight <1500 g.5 Impaired cognitive function and lower academic performance at school age seem to persist into young adulthood,5 although a majority seem to have overcome earlier difficulties at the age of 22 to 25 years.6
Despite the abundance of data regarding test results and disorders of extremely preterm children, the overall importance of these minor and major disabilities for their educational and vocational career as adults is yet to be described. Also, few follow-up studies have addressed the situation of the much larger group of children born moderately preterm.7,8 In this study, we used the Swedish national databases to study social outcomes in an entire national cohort of young adults, 23 to 29 years of age, in relation to gestational age.
This study was based on Swedish national registers held by the National Board of Health and Welfare and Statistics Sweden linked through each individual's unique personal identification number. A total of 570768 liveborn individuals were identified in the Swedish medical birth register 1973–1979 with a reported gestational age ≤41 weeks.9 In this register age, we also identified the personal identification number of the mother, geographic location of the household, birth date and gender of the child, and major perinatal indicators (gestational age, Apgar <7 at 5 minutes, neonatal distress, birth weight, small for gestational age [SGA], preeclampsia, abruptio placentae, preterm rupture of the membranes). Gestational age was defined according to maternal report of last menstrual period and clinical judgment by the attending pediatrician. SGA was defined as less than −2 SD according to the growth chart developed by Marsal et al.10 A total of 4022 (0.7%) children with a registered birth weight >3 SD or less than −6 SD according to this same growth chart were excluded from the study population as probable coding errors, because these errors are more probable with birth weights categorized as large for gestational age than in children categorized as SGA in preterm infants. A total of 12927 (2.3%) individuals with at least 1 reported malformation other than undescended testicle, preauricular appendage, congenital nevus, or hip dislocation were also excluded, as were 7310 (1.3%) individuals who were reported dead in the Swedish national cause-of-death register by 2002. The infant (1 year) mortality according to gestational age was 43.4% in weeks 24 to 28, 13.6% in weeks 29 to 32, 1.6% in weeks 33 to 36, 0.38% in weeks 37 to 38, and 0.17% in weeks 39 to 41 when infants with major malformations were excluded. Finally, we excluded 27041 persons (4.7%) who were no longer residents in Sweden in 2002 according to the register of the total population in December 2002, leaving 522310 individuals with a gestational age between 24 and 41 weeks to be included in the study population
The socioeconomic status (SES) of the household, housing situation, maternal country of birth, and lone parent households were identified in the Swedish Population and Housing Census of 1985. Socioeconomic groups were defined according to a classification created by Statistics Sweden (manual workers, skilled workers, and white collar 1–3). This hierarchical classification of occupation is based on the educational level required but also takes type of work and the position at the work place into consideration. Farmers, self-employed and unemployed, were categorized as unclassified. The head of the household was defined as the adult in the household with the highest SES.11 Maternal country of birth was categorized into 4 geographical groups: Sweden, Finland, the rest of Europe, and the rest of the world. Social welfare benefits received by the household of the mother were added through linkage to the Total Enumeration Income Survey for 1990.
Several social outcome variables were created with 2002 information from the Total Enumeration Income Survey that year: (1) sickness pension, indicating lifelong pension because of long-standing illness or disability, (2) handicap allowance, indicating a permanent disability, (3) employment, indicating having an income from employment or own firm in November 2002, (4) illness benefits, indicating temporary economic support during at least 2 consecutive weeks because of illness from the national health insurance, (5) residence in the household of the biological parents, (6) student, as indicated by having received student support or loan, (7) net salary, indicating net income from employment and own firm, (8) disposable income, indicating the sum of all incomes, including societal benefits, deducted by income tax, and (9) net transfer to society, as indicated by deducting disposable income from work income. Economic compensation because of disability assistance, indicating the need for a personal helper at least 4 hours daily, was identified in the register kept at the National Social Insurance Board.
The number of months that social assistance was received during 2002 was identified in the Swedish social-assistance register, and the highest completed education as of December 2002 was derived from the Swedish educational register. Education was categorized as “basic” if the study participant had completed no more than the compulsory 9 years of primary school, and as “postsecondary” if ≥1 educational level had been completed after secondary school. A summarized disability outcome variable was created that indicated having received sickness pension, handicap allowance, and/or disability assistance.
Multivariate analyses of the dichotomised social outcome variables described above were calculated by logistic regression on the log scale to calculate estimates equivalent to relative risk ratios (RRs).12 We calculated 95% confidence intervals by using the test-based method. Disability and residence in the household of the biological parents were analyzed in the entire study population. Disabled individuals, however, were excluded from the analysis of educational outcomes to allow for an analysis of more subtle consequences. Employment, finally, was studied in those in the study population that had neither an indication of disability nor of being a student.
Year of birth was entered as a continuous variable in the regression models. Missing data were entered as a separate category in the analytic models. Model 1 in the logistic regression analysis was adjusted for gender and age only. Socioeconomic and parental morbidity variables were added as confounders in model 2. SGA and multiple birth were added to the variables of model 2 as perinatal confounders in model 3. Interaction effects were tested in logistic regression models. Attributable risk was calculated according to the formula suggested by Rothman and Greenland.13
Income variables were analyzed in a linear regression with the same independent variables as in the logistic regression models described above. Students were excluded from this analysis, and only those employed without a disability were included in the analysis of net salary. All statistical analyses were conducted by using SAS 9.0 software (SAS Institute, Inc, Cary, NC).
There were 431656 individuals who were born term (39–41 weeks' gestational age), 68541 born slightly preterm (37–38 weeks' gestational age), 19166 born moderately preterm (33–36 weeks' gestational age), 2947 born very preterm (317 born 24–28 weeks' gestational age, and 2630 born 29–32 weeks’ gestational age) in the study population. Among the very preterm infants, 43 were born with a gestational age of ≤26 weeks.
Sociodemographic data of the study population by gestational age are presented in Table 1. Boys were more often born very (24–32 weeks' gestational age) and moderately preterm (33–36 weeks' gestational age). In 1985, the households of the study subjects born preterm more often lived in rented apartments, received social welfare, and had an SES as manual workers compared with the households of those born at term.
Perinatal variables by gestational age are presented in Table 2. Perinatal and gestational complications increased by increasing degree of preterm birth.
The rate of disability in the entire study population decreased by year of birth from 1.66% in those born 1973 to 1.46% in those born 1979; a RR of 0.97 (95% confidence interval [CI]: 0.96–0.98) per year. Social outcomes in relation to gestational age at birth are presented in Table 3. Rates of disability allowance, sickness pension, and disability assistant increased gradually in step with the degree of preterm birth. More than 85% of even the extremely preterm children (24–28 weeks' gestation), however, had no such indication of a disability. After adjustment for age and gender in model 1 in the multivariate analysis, the RRs of children born very preterm was 4.39 (Table 4) for having ≥1 indication of a disability. Adding socioeconomic and perinatal confounders decreased the RRs slightly to 3.76. Moderately preterm (33–36 weeks' gestational age) and slightly preterm (37–38 weeks' gestational age) had RRs of 1.51 and 1.23, respectively, in the final model. The total attributable risk for disability associated with preterm births was 7.0%, of which 5.2% was accounted for by moderately and slightly preterm births. Having received social welfare benefits and sick allowance had a distribution by gestational age that was similar to the disability indicators.
Study subjects born preterm more often resided in the household of their parents compared with peers of the same gender and age who were born term (RR: 1.15 [95% CI: 1.05–1.25], 1.05 [95% CI: 1.01–1.09], and 1.03 [95% CI: 1.01–1.05] for extremely, moderately, and slightly preterm groups, respectively [data not shown in the tables]). Adding confounders only marginally changed these estimates.
A total of 26% of those born with a gestational age of 28 weeks or less had a university education in comparison with 38% of those with a normal gestational age (39–41 weeks). When disabled individuals were excluded, this difference decreased to an adjusted RR of 0.93 (Table 4) for having a university education. Moderately preterm individuals had a RR of 0.96 in the final model, whereas slightly preterm individuals had a similar chance as those born term to have a university education in the final model (Table 4).
A total of 74.1% of the study population born at term had an income from employment in November 2002 compared with 68.1% of those born in gestational weeks 24 to 28, 70.1% in weeks 29 to 32, and 72.5% of those born in weeks 33 to 36. When disabled individuals were excluded, however, preterm and term individuals had very similar chances of having being employed (Table 4). As Table 5 demonstrates, very preterm birth was associated with a lower net salary among those who were employed in a linear regression analysis of income, even after disabled individuals were excluded and socioeconomic and perinatal confounders were accounted for. Adults born preterm also had a lower disposable income than adults born term (Table 5). When the disposable income was subtracted from the net salary, a net mean individual transfer of 3079 euros to the society from the individuals born term in the study population was identified in 2002 (Table 5). In a linear regression, it was demonstrated that this transfer was 715 euros lower (P = .02) in those born very preterm and 171 euros lower (P = .15) in those born moderately preterm after socioeconomic and perinatal confounders were accounted for (Table 5). In the entire population, this amounts to a net negative transfer of ∼65 million euros from individuals born preterm compared with those born term in 2002 alone, with those born moderately preterm accounting for 77% of that amount.
Differences in effects of preterm birth of gender and SES on university education, employment, and disability were studied in interaction analyses with the variables in model 1 and a dichotomized SES variable in the logistic regression analysis presented above. The effect of preterm birth on the chance of receiving a university education was greater in families with a low SES compared with a high SES (P = .005); otherwise, no statistically significant interaction effects for SES or gender were found.
In this national cohort study of Swedish residents born in the 1970s, we have demonstrated that preterm birth is associated with educational and vocational impairment in young adulthood. As expected, the effect was greatest for the 0.6% born very preterm (24–32 weeks' gestational age), but even in this group, the large majority were employed and contributed more in income tax than they received in benefits. Somewhat more unexpected were the considerable effects observed in the 3.7% who were born moderately preterm.
Preterm birth was associated with a stepwise increased risk of disability but also with a lower net income and a lower chance of completing a university education than in those without any indication of a disability. A number of obstetrical and neonatal complications, such as intrauterine growth retardation, cerebral hemorrhage and infarction, hypoglycemia, pulmonary disease, septicemia, and asphyxia, may cause long-term disability through injuries to the immature central nervous system and are more common in preterm births. It seems less certain that these catastrophic injuries are responsible for the more subtle educational and vocational impairment of preterm children without an identified disability. It is possible that mechanisms such as incomplete myelinization14,15 and poor growth of the gray matter of the preterm brain16–18 have to be considered to explain these effects.
This study demonstrates an attributable risk of 7% for preterm birth on disability in this cohort of young adults, in which moderately and slightly preterm births accounted for 5%. In a similar manner, moderately preterm accounted for 61% of the societal transfers to individuals born preterm. These findings are in accord with recent studies that have described considerable mortality and respiratory morbidity in infants born at 30 to 34 weeks' gestation,8 as well as in those with 35 to 36 weeks' gestation.8 The results of this study support calls for more research and development of improved secondary preventive strategies in this group of infants.
The analysis of income and societal transfers in this study indicates that the Swedish society as a whole has gained considerably from the increased survival and productivity in vulnerable neonates born preterm associated with the development of neonatal intensive care. Even very preterm infants made such economic contributions to the society in young adulthood (mean: 2000 euros/year) that it seems most probable that this would hold true even if other societal expenses, such as increased needs for medical care and drugs, were added to the bill. Compared with children born term, however, children born preterm contributed less to society. This demonstrates that even quite ambitious and costly primary and secondary preventive programs in perinatal health have a potential of economic gains for the society in the long run. The potential economic gain for the Swedish society, in the cohorts in this study if all long-term effects of preterm birth could have been prevented would have been ∼65 million euros in 1 year alone in terms of taxes and decreased costs for benefits.
The major strength of this study is the large study population made possible by the high quality and extensive coverage of the Swedish national registers, which enabled us to study a large number of infants with marginal attrition. The use of register data on income, disability, and education excludes the possibility of self-reporting bias, which is otherwise a common problem in follow-up studies We were able to include a wide range of socioeconomic, perinatal, and parental morbidity variables as possible confounding factors in the analysis by linking several national registers. It is still quite probable, however, that important factors associated with preterm birth that might also affect the fetus directly, such as viral infections or toxic agents, were not accounted for in the multivariate analysis.
The cohort in this study was born before ultrasound became a routine procedure to measure gestational age in early pregnancy. Although a certain socioeconomic bias in coding may be inherent with the variable of gestational age used in this study, most of this bias can be expected to be taken care of in the multivariate analysis. Coding errors of gestational age tend to create outliers that are falsely labeled preterm births.19 We excluded individuals with extremely uncommon combinations of birth weight and gestational age to minimize this problem, but it seems likely that some miscoded individuals born at term remained in the study population coded as preterm, thus tending to attenuate the true associations of negative outcomes with preterm birth.
A recent analysis of cerebral palsy in neonates with a very low birth weight in 16 different European cities demonstrates decreasing rates in 1996 compared with 1980 despite higher survival rates of infants with very low birth weight in 1996.20 The lower rate of cerebral palsy was particularly noted among children born in gestational weeks 29 to 32, whereas the rate remained quite similar over time in children born at 24 to 28 weeks' gestation. Swedish studies have demonstrated a similar pattern of improvement.21 Thus, we may expect that disability rates of preterm with a gestation of 29 to 32 weeks born in Sweden during more recent years will have even more favorable outcomes than those reported in this study of cohorts from the 1970s.
This study demonstrates that there is considerable long-term impairment associated with preterm birth. This impairment generates great costs for the society, although the large majority of even the most immature children seemed to live a productive and self-supported life in early adulthood. More attention is needed toward the moderately preterm children who accounted for most of the chronic disability associated with preterm birth in this population of young adults.
This work was supported by the National Board of Health and Welfare, Stockholm, Sweden.
- Accepted March 2, 2007.
- Address correspondence to Anders Hjern, MD, PhD, Center for Epidemiology, National Board of Health and Welfare, 106 30 Stockholm, Sweden. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Escobar GJ, McCormick MC, Zupancic JA, et al. Unstudied infants: outcomes of moderately premature infants in the NICU. Arch Dis Child Fetal Neonatal Ed.2006;91 :F238– F44
- ↵Centre for Epidemiology. The Swedish Medical Birth Registry. A summary of content and quality. Stockholm, Sweden: National Board of Health and Welfare; 2003. Report 2003-112-3
- ↵Statistics Sweden. Socio-economic Classification (SEI). Stockholm, Sweden: Statistics Sweden; 1982
- ↵Rothman K, Greenland S. Modern Epidemiology. 2nd ed. Philadelphia, PA: Lippincott-Raven Publishers; 1998
- ↵Volpe JJ. Cerebral white matter injury of the premature infant: more common than you think. Pediatrics.2003;112 :176– 180
- Peterson BS, Anderson AW, Ehrenkranz R, et al. Regional brain volumes and their later neurodevelopmental correlates in term and preterm infants. Pediatrics.2003;111 :939– 948
- ↵Murphy BP, Inder TE, Huppi PS, et al. Impaired cerebral cortical gray matter growth after treatment with dexamethasone for neonatal chronic lung disease. Pediatrics.2001;107 :217– 221
- ↵Haglund B. Birthweight distributions by gestational age: comparison of LMP-based and ultrasound-based estimates of gestational age using data from the Swedish Birth Registry. Paediatr Perinat Epidemiol.2007: In press
- Copyright © 2007 by the American Academy of Pediatrics