Gestational Age at Birth and Mortality in Young Adulthood

Context Preterm birth is the leading cause of infant mortality in developed countries, but the association between gestational age at birth and mortality in adulthood remains unknown.

Objective To examine the association between gestational age at birth and mortality in young adulthood.

Design, Setting, and Participants National cohort study of 674 820 individuals born as singletons in Sweden in 1973 through 1979 who survived to age 1 year, including 27 979 born preterm (gestational age <37 weeks), followed up to 2008 (ages 29-36 years).

Main Outcome Measures All-cause and cause-specific mortality.

Results A total of 7095 deaths occurred in 20.8 million person-years of follow-up. Among individuals still alive at the beginning of each age range, a strong inverse association was found between gestational age at birth and mortality in early childhood (ages 1-5 years: adjusted hazard ratio [aHR] for each additional week of gestation, 0.92; 95% CI, 0.89-0.94; P < .001), which disappeared in late childhood (ages 6-12 years: aHR, 0.99; 95% CI, 0.95-1.03; P = .61) and adolescence (ages 13-17 years: aHR, 0.99; 95% CI, 0.95-1.03; P = .64) and then reappeared in young adulthood (ages 18-36 years: aHR, 0.96; 95% CI, 0.94-0.97; P < .001). In young adulthood, mortality rates (per 1000 person-years) by gestational age at birth were 0.94 for 22 to 27 weeks, 0.86 for 28 to 33 weeks, 0.65 for 34 to 36 weeks, 0.46 for 37 to 42 weeks (full-term), and 0.54 for 43 or more weeks. Preterm birth was associated with increased mortality in young adulthood even among individuals born late preterm (34-36 weeks, aHR, 1.31; 95% CI, 1.13-1.50; P < .001), relative to those born full-term. In young adulthood, gestational age at birth had the strongest inverse association with mortality from congenital anomalies and respiratory, endocrine, and cardiovascular disorders and was not associated with mortality from neurological disorders, cancer, or injury.

Conclusion After excluding earlier deaths, low gestational age at birth was independently associated with increased mortality in early childhood and young adulthood.

Preterm birth, defined as birth that occurs prior to 37 completed weeks of gestation, is the leading cause of perinatal morbidity and mortality in developed countries.¹ Although the early effects of preterm birth are well documented,² less is known about the longer-term outcomes in adulthood. These outcomes have a growing clinical and public health importance because of the high prevalence of preterm birth and improved early survival. In the past 3 decades, the prevalence of preterm birth in the United States has increased to more than 12%,³ similar to the prevalence in Africa and south Asia,⁴ compared with 5% to 9% in Europe.¹ In the same period, advances in neonatal care, with widespread use of antenatal corticosteroids, surfactant therapy, and high-frequency ventilation, have led to improved survival of preterm infants.² As a result, large numbers of individuals who were born preterm are now surviving to adulthood. A comprehensive understanding of their outcomes in adulthood is needed to enable earlier prevention, detection, and treatment of the long-term health sequelae.

Earlier studies have explored the relationship between preterm birth and mortality in childhood and adolescence⁵ or between low birth weight and mortality in adulthood.⁶ To our knowledge, no studies to date have reported the specific contribution of gestational age at birth on mortality in adulthood. The limitations of focusing on birth weight, which is a function of gestational age and fetal growth, have been well documented.^7,8 The relative influences of gestational age and fetal growth are important to disentangle because they have different underlying mechanisms that may require different preventive interventions.

We conducted a national cohort study in Sweden to examine the association between gestational age at birth, independent of fetal growth, and mortality in young adulthood. A national cohort of singleton infants born in 1973 through 1979 was followed up into young adulthood for all-cause and cause-specific mortality. We hypothesized that low gestational age at birth would be independently associated with increased mortality in young adulthood among individuals who were still alive at the beginning of this age range.

To examine long-term mortality after infancy, we identified 678 528 individuals from the Swedish Birth Registry who were born as singletons from 1973 through 1979 and who survived to age 1 year. We excluded 2756 individuals (0.4%) who had missing information for gestational age at birth and 248 others (<0.1%) who had missing information for birth weight. To remove possible coding errors, we excluded 704 individuals (0.1%) who had a reported birth weight more than 4 SDs above or below the mean birth weight for gestational age and sex from a Swedish reference growth curve.⁹ A total of 674 820 individuals (99.5% of the original cohort) remained for inclusion in the study. This study was approved by the ethics committee of Lund University in Malmö, Sweden. Informed consent was waived as a requirement by the ethics committee.

The study cohort was followed for all-cause and cause-specific mortality from age 1 year to the maximum possible follow-up (ages 29-36 years). Deaths were identified through December 31, 2008, using the Swedish Death Registry. This registry includes information on all deaths in Sweden with compulsory reporting nationwide. Cause of death is classified according to the International Classification of Diseases (ICD) (revisions 8, 9, and 10).

The predictor of interest was gestational age at birth, which was based on maternal report of last menstrual period and identified using prenatal and birth records in the Swedish Birth Registry. Gestational age at birth was analyzed alternatively as a continuous variable (in weeks) and as a categorical variable in 5 groups (22-27, 28-33, 34-36, 37-42, and ≥43 weeks) to allow for a nonlinear response. Completeness of the birth and death registries is nearly 100%.

Other perinatal and demographic characteristics that may be associated with preterm birth and mortality were identified using the Swedish Birth Registry and national census data, which were linked using an anonymous personal identification number. The following variables were included as model covariates: sex, birth year (as a continuous variable), maternal age at birth (<20, 20-24, 25-29, 30-34, or ≥35 years), and maternal marital status (married/cohabiting, never married, divorced/widowed, or unknown). Education was also entered in the model separately for mothers and fathers as one of the following: compulsory high school or less (≤9 years); practical high school or some theoretical high school (10-11 years); theoretical high school, college, or both (≥12 years); or unknown.

Also included as a model covariate was fetal growth measured as the number of standard deviations from the mean birth weight for gestational age and sex from a Swedish reference growth curve⁹ and categorized into 6 groups (<−2 SD; −2 to <−1 SD; −1 to <0 SD; 0 to <1 SD; 1 to <2 SD; or ≥2 SD) to allow for a nonlinear response. Birth order was included as 1, 2, or 3 or greater because low birth order has been associated with preterm birth¹⁰ and high birth order may be associated with increased mortality in adulthood.¹¹

Cox proportional hazards regression was used to estimate hazard ratios (HRs) and 95% CIs for the association between gestational age at birth and mortality in early childhood (ages 1-5 years), late childhood (ages 6-12 years), adolescence (ages 13-17 years), and young adulthood (age 18 years to the maximum follow-up of 29-36 years). This association was estimated in each of these age ranges after excluding earlier deaths, ie, among the total number of individuals who were still alive at the beginning of each age range. Individuals were censored at year of emigration (n = 36 065; 5.3%), which was determined by the absence of a Swedish residential address in census data. Individuals who emigrated had a similar gestational duration compared with those who did not emigrate (mean, 40.2 vs 40.4 weeks; P = .72); thus, it is unlikely that emigration introduced any substantial bias.

Analyses were conducted first unadjusted and then adjusted for covariates. Likelihood ratio tests were used to test for a trend of survival functions across gestational age at birth (as an ordered categorical variable) and to test for departure from linear trend. The same analyses were repeated to examine cause-specific mortality in each age range for all ICD categories with at least 100 total deaths. The proportional hazards assumption was tested for each model using the method described by Grambsch and Therneau.¹² First-order interactions were explored between gestational age and the model covariates with respect to mortality, using a likelihood ratio test to evaluate for statistical significance. All statistical tests were 2-sided and used an α level of .05.

We also performed 3 sensitivity analyses. First, we assessed the sensitivity of results to congenital malformations by repeating the main analyses after excluding individuals with congenital malformations identified in the birth records or as a cause of death (n = 8641; 1.3%). Second, we assessed the sensitivity of results to injury deaths by repeating the analyses after excluding all 1617 deaths due to injury (ie, censoring these individuals at the time of death). Third, we explored for potential bias due to missing data for gestational age (n = 2756; 0.4%) and birth weight (n = 248; <0.1%) by repeating the analyses using a standard multiple imputation procedure for these missing data.^13,14 All analyses were conducted using Stata version 11.0 (StataCorp, College Station, Texas).¹³

The total prevalence of preterm birth in Sweden during 1973-1979 was 5.0%. The proportion of infants born preterm (<37 weeks) who died in the first year of life was 8.6%, including 70.7% of those born at 22 to 27 weeks, 18.4% of those born at 28 to 33 weeks, and 3.1% of those born at 34 to 36 weeks, compared with 0.7% of those born full-term (37-42 weeks) and 1.0% of those born postterm (≥43 weeks).

All subsequent analyses were based on all 674 820 individuals born as singletons during 1973-1979 who survived to age 1 year. Of this number, 27 979 (4.1%) were born preterm (<37 weeks), including 226 (0.03%) born at 22 to 27 weeks, 5163 (0.8%) born at 28 to 33 weeks, and 22 590 (3.3%) born at 34 to 36 weeks. Preterm infants were more likely than full-term infants to be male or first-born; their mothers were more likely to be either younger than 20 years or 35 years or older at the time of delivery or to be unmarried; and their mothers and fathers were more likely to have low or unknown educational attainment (Table 1).

A total of 7095 deaths occurred in 20.8 million person-years of follow-up from age 1 year to the maximum attained ages of 29 to 36 years. The overall mortality rates per 1000 person-years were 0.33 in early childhood (ages 1-5 years), 0.15 in late childhood (ages 6-12 years), 0.25 in adolescence (ages 13-17 years), and 0.47 in young adulthood (ages 18-36 years).

Among individuals still at risk in each time period, a strong inverse association was observed between gestational age at birth and mortality in early childhood (adjusted hazard ratio [aHR] for each additional week of gestation, 0.92; 95% CI, 0.89-0.94; P < .001), no association was observed in late childhood (aHR, 0.99; 95% CI, 0.95-1.03; P = .61) or adolescence (aHR, 0.99; 95% CI, 0.95-1.03; P = .64), and an inverse association reappeared in young adulthood (aHR, 0.96; 95% CI, 0.94-0.97; P < .001) (Table 2). In early childhood and young adulthood, the inverse association between gestational age at birth and mortality was consistent with a linear relationship, and tests for departure from linearity were nonsignificant. Adjustment for all or any subset of covariates had only a modest effect on any of the risk estimates. These associations also did not vary by sex in any age range in stratified models, and the interaction between gestational age (modeled as either a continuous or categorical variable) and sex was nonsignificant; hence, the results are reported unstratified (Table 2).

In early childhood as well as young adulthood, preterm birth was associated with increased mortality even among individuals born late preterm (34-36 weeks). Mortality rates per 1000 person-years for individuals born late preterm and full-term, respectively, were 0.53 and 0.32 in early childhood and 0.65 and 0.46 in young adulthood. Adjusted hazard ratios for the association between late preterm birth (34-36 weeks) and mortality were 1.53 (95% CI, 1.18-2.00; P = .001) in early childhood and 1.31 (95% CI, 1.13-1.50; P < .001) in young adulthood, relative to individuals born full-term.

In contrast, gestational age at birth was not associated with mortality in late childhood (ages 6-12 years) or adolescence (ages 13-17 years). The association between late preterm birth and mortality in these age ranges had a modestly increased point estimate, but it did not reach statistical significance (Table 2). Combining these age groups to improve statistical power did not appreciably alter the risk estimates and they remained nonsignificant.

Risk estimates for the association between model covariates and all-cause mortality in young adulthood (ages 18-36 years) are presented in Table 3. After adjusting for the other variables included in the model, mortality was positively associated with male sex, birth order of 3 or more, and having an unmarried mother and was inversely associated with maternal age and maternal and paternal education level. Low fetal growth (<−2 SD or −2 to <−1 SD from the reference birth weight for gestational age and sex) was also associated with increased mortality in young adulthood after adjusting for gestational age and the other covariates. The relationship between fetal growth and mortality, however, was not linear (P for departure from linear trend = .006). We found no first-order interactions between gestational age at birth and the covariates, including sex, with respect to mortality.

Cause-specific mortality is presented in Table 4. In early childhood (1-5 years), gestational age at birth had the strongest inverse association with mortality from congenital anomalies (which were mostly heart defects) (aHR for each additional week of gestation, 0.83; 95% CI, 0.79-0.87) and endocrine and respiratory disorders. There were no significant associations with mortality from cardiovascular disease, neurological disorders, cancer, or injury. In late childhood (ages 6-12 years) and adolescence (ages 13-17 years), whether modeled as separate age groups or combined, there were no significant associations between gestational age at birth and any cause-specific mortality.

In young adulthood (ages 18-36 years), gestational age at birth had the strongest inverse associations with mortality from congenital anomalies (aHR for each additional week of gestation, 0.80; 95% CI, 0.74-0.86), respiratory disorders (aHR, 0.85; 95% CI, 0.76-0.94), and endocrine disorders (aHR, 0.88; 95% CI, 0.81-0.97). An inverse association with cardiovascular mortality also emerged in this age range (aHR, 0.93; 95% CI, 0.88-0.99). In contrast, gestational age at birth (modeled as either a continuous or categorical variable) was not significantly associated with mortality from neurological disorders, cancer, or injury in young adulthood.

Exclusion of individuals with congenital malformations affected the results only in early childhood (ages 1-5 years). The inverse association between gestational age at birth and mortality in this age range was attenuated although it remained significant (aHR for each additional week of gestation, 0.96; 95% CI, 0.93-0.99; P = .03). The risk estimates for all other age ranges remained virtually identical (eTable 1).

Exclusion of deaths due to injury also affected the results only in early childhood (ages 1-5 years). The inverse association between gestational age at birth and mortality in this age range was slightly strengthened (aHR for each additional week of gestation, 0.88; 95% CI, 0.85-0.91; P < .001), whereas the risk estimates for all other age ranges remained virtually identical (eTable 2).

Infants with missing data for gestational age (n = 2756; 0.4%) had a mean birth weight (3339 g) and mortality rate (0.43/1000 person-years) that were intermediate to those of late preterm infants (2743 g; 0.48/1000 person-years) and full-term infants (3536 g; 0.33/1000 person-years). Using multiple imputation of these missing data,^13,14 the risk estimates and significance levels in all age ranges were negligibly affected, suggesting that the relatively small amount of missing data did not introduce any appreciable bias.

We found that low gestational age at birth was independently associated with increased mortality in young adulthood among individuals born in Sweden in 1973-1979. This association was consistent with a linear relationship and was independent of fetal growth as well as other perinatal and socioeconomic factors. Preterm birth was associated with increased mortality in young adulthood even among those born late preterm (34-36 weeks). Across a longer age range from 1 year to 36 years, a strong inverse association between gestational age at birth and mortality was observed in early childhood, which disappeared in late childhood and adolescence and then reappeared in young adulthood.

To our knowledge, this is the first study to report the specific contribution of gestational age at birth on mortality in adulthood. The results underscore the persistent long-term health sequelae of preterm birth. Earlier studies have explored the association between low birth weight and mortality in adulthood without examining the specific contribution of gestational age.^15-29 A recent meta-analysis of these studies reported an inverse association between birth weight and mortality in adulthood, with a 6% lower risk per kilogram of higher birth weight (aHR, 0.94; 95% CI, 0.92-0.97).⁶ Another study in Norway reported that preterm birth was associated with increased mortality in early and late childhood but not in adolescence.⁵ That study, however, did not adjust for fetal growth and did not report mortality in adulthood. The relative influences of gestational age and fetal growth are difficult to disentangle completely because of correlation, but our findings suggest that preterm birth is an important independent risk factor for mortality in young adulthood. Furthermore, although men have increased mortality rates overall, the association we found between gestational age at birth and mortality affected both sexes similarly, without evidence of effect modification.

We found multiple causes for this association, especially congenital anomalies, respiratory and endocrine disorders, and cardiovascular disorders in young adulthood. A number of previous studies have reported an inverse association between birth weight and mortality from cardiovascular disease,^6,20,30 whereas data on gestational age at birth and other cause-specific mortality are scarce. Our cause-specific findings are generally consistent with associations we previously reported between low gestational age and various morbidities in the same cohort, including asthma,³¹ hypertension,³² diabetes,³³ and hypothyroidism.³⁴ The underlying mechanisms are still largely unknown but may involve a complex interplay of fetal and postnatal nutritional abnormalities^35,36; other intrauterine exposures, including glucocorticoid and sex hormone alterations³⁷; and common genetic factors.³⁸

Our findings were not adequately explained by congenital malformations, despite their strong association with preterm birth. Cardiovascular and other malformations are more than twice as common among preterm than full-term infants^39-41 and have a high mortality rate.^41-43 However, we found that after excluding congenital malformations, the inverse association between gestational age at birth and mortality was weakened but persisted in early childhood, and was virtually unaffected in young adulthood.

The prevalence of preterm birth currently exceeds 12% in the United States,³ more than double the prevalence observed in this earlier Swedish cohort. Despite considerable research and public health efforts, preterm birth has increased in the United States³ and in Europe⁴ during the past 3 decades. Much of the increase is due to rising numbers of medically indicated preterm births, as well as increased use of assisted reproductive technologies.⁴⁴ More than 520 000 births now occur prematurely every year in the United States,³ costing more than $26 billion annually in health care expenditures and lost productivity.⁴⁵ Although most survivors have a high level of function and self-reported quality of life in young adulthood,² our previous^31-34,46 and current findings demonstrate the increased long-term morbidities and mortality that may also be expected. Clinicians will increasingly encounter the health sequelae of preterm birth throughout the life course and will need to be aware of the long-term effects on the survivors, their families, and society.²

The most important strength of the current study was its ability to examine the association between gestational age at birth and mortality in a large national cohort that was followed up into young adulthood, using birth and death registry data that are nearly 100% complete. The results were adjusted for fetal growth as well as other broadly measured potential confounders.

Limitations included the estimation of gestational age by maternal report of last menstrual period rather than by ultrasound, which was not yet widely used at the time these individuals were born (1973-1979). Statistical power was limited to detect associations with specific causes of mortality. Information was unavailable for mode of delivery, postnatal growth patterns, and smoking history. Cigarette smoking has a reported prevalence of 26% among women and 22% among men in Sweden,⁴⁷ is a risk factor for preterm birth,⁴⁸ and may have disproportionate health effects among survivors of preterm birth.⁴⁹ Finally, preterm infants today may have different long-term outcomes from this cohort due to advances in neonatal care and improved survival at earlier gestational ages. It is unclear to what extent our findings are generalizable to later cohorts, and any such comparisons should be made with caution. The relationships between neonatal care and preterm birth sequelae are highly dynamic and will require additional follow-up of other large cohorts in the future.

In summary, among individuals born in Sweden in 1973-1979, low gestational age at birth was independently associated with increased mortality in early childhood and young adulthood. These findings underscore the need for more effective strategies to prevent preterm birth, better understanding of causal pathways, and increased awareness of the long-term health sequelae throughout the life course.

Corresponding Author: Casey Crump, MD, PhD, Department of Medicine, Stanford University, 900 Blake Wilbur Dr, Stanford, CA 94304-2205 (kccrump@stanford.edu).

Author Contributions: Dr J. Sundquist had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Crump, K. Sundquist, J. Sundquist, Winkleby.

Acquisition of data: J. Sundquist.

Analysis and interpretation of data: Crump, K. Sundquist, J. Sundquist, Winkleby.

Drafting of the manuscript: Crump, Winkleby.

Critical revision of the manuscript for important intellectual content: Crump, K. Sundquist, J. Sundquist, Winkleby.

Statistical analysis: Crump, J. Sundquist, Winkleby.

Obtained funding: K. Sundquist, J. Sundquist.

Administrative, technical, or material support: J. Sundquist.

Study supervision: J. Sundquist, Winkleby.

Funding/Support: This study was supported by grants from the National Institute of Child Health and Human Development (1R01HD052848-01), the Swedish Research Council (2008-3110 and 2008-2638), the Swedish Council for Working Life and Social Research (2006-0386, 2007-1754, and 2007-1962), and an ALF project grant (Avtal om Läkarutbildning och Forskning, or Agreement on Medical Training and Research), Lund, Sweden.

Role of the Sponsors: The funding agencies had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.