Association Between Maternal Body Mass Index in Early Pregnancy and Incidence of Cerebral Palsy

Key Points

Question  Is maternal obesity in early pregnancy associated with incidence of cerebral palsy in the offspring regardless of gestational age at delivery?

Findings  In this nationwide cohort study of 1.4 million singleton Swedish children, maternal overweight and obesity were statistically significantly associated with increased rates of cerebral palsy. The association was restricted to children born at full term and partly mediated through asphyxia-related neonatal complications.

Meaning  Maternal obesity was associated with an increased rate of cerebral palsy in the offspring, partly mediated by birth asphyxia.

Importance  Maternal overweight and obesity are associated with increased risks of preterm delivery, asphyxia-related neonatal complications, and congenital malformations, which in turn are associated with increased risks of cerebral palsy. It is uncertain whether risk of cerebral palsy in offspring increases with maternal overweight and obesity severity and what could be possible mechanisms.

Objective  To study the associations between early pregnancy body mass index (BMI) and rates of cerebral palsy by gestational age and to identify potential mediators of these associations.

Design, Setting, and Participants  Population-based retrospective cohort study of women with singleton children born in Sweden from 1997 through 2011. Using national registries, children were followed for a cerebral palsy diagnosis through 2012.

Exposures  Early pregnancy BMI.

Main Outcomes and Measures  Incidence rates of cerebral palsy and hazard ratios (HRs) with 95% CIs, adjusted for maternal age, country of origin, education level, cohabitation with a partner, height, smoking during pregnancy, and year of delivery.

Results  Of 1 423 929 children included (mean gestational age, 39.8 weeks [SD, 1.8]; 51.4% male), 3029 were diagnosed with cerebral palsy over a median 7.8 years of follow-up (risk, 2.13 per 1000 live births; rate, 2.63/10 000 child-years). The percentages of mothers in BMI categories were 2.4% at BMI less than 18.5 (underweight), 61.8% at BMI of 18.5 to 24.9 (normal weight), 24.8% at BMI of 25 to 29.9 (overweight), 7.8% at BMI of 30 to 34.9 (obesity grade 1), 2.4% at BMI of 35 to 39.9 (obesity grade 2), and 0.8% at BMI 40 or greater (obesity grade 3). The number of cerebral palsy cases in each BMI category was 64, 1487, 728, 239, 88, and 38; and the rates per 10 000 child-years were 2.58, 2.35, 2.92, 3.15, 4.00, and 5.19, respectively. Compared with children of normal-weight mothers, adjusted HR of cerebral palsy were 1.22 (95% CI, 1.11-1.33) for overweight, 1.28 (95% CI, 1.11-1.47) for obesity grade 1, 1.54 (95% CI, 1.24, 1.93) for obesity grade 2, and 2.02 (95% CI, 1.46-2.79) for obesity grade 3. Results were statistically significant for children born at full term, who comprised 71% of all children with cerebral palsy, but not for preterm infants. An estimated 45% of the association between maternal BMI and rates of cerebral palsy in full-term children was mediated through asphyxia-related neonatal morbidity.

Conclusions and Relevance  Among Swedish women with singleton children, maternal overweight and obesity were significantly associated with the rate of cerebral palsy. The association was limited to children born at full term and was partly mediated through asphyxia-related neonatal complications.

Quiz Ref IDCerebral palsy is the most common pediatric motor disability. In the United States, the estimated prevalence by age 8 years was 3.1 per 1000 live births in 2008,¹ and in Sweden, 2.2 per 1000 live births in 2006.² Despite advances in obstetric and neonatal care, cerebral palsy prevalence increased from 1998 through 2006 in children born at full term.² Cerebral palsy constitutes a significant lifetime disability for children and their families, and is associated with chronic diseases³ and a shortened life expectancy.⁴

Few preventable factors are known to affect the risk of cerebral palsy. Although preterm delivery is a risk factor, most children with cerebral palsy are born at full term and the causes of cerebral palsy in this group remain largely unknown.⁵ Birth asphyxia has been implicated, but its contribution relative to other perinatal factors is controversial.⁵

Quiz Ref IDMaternal obesity is a major prenatal risk factor for obstetric complications,⁶ preterm delivery,⁷ asphyxia-related neonatal morbidities,⁸ and possibly cognitive and behavioral developmental disabilities in children.⁹ An association between high maternal body mass index (BMI) and cerebral palsy has been suggested^10-14 but never addressed in a nationwide investigation with sufficiently large numbers of cases to allow exploration of possible causal pathways.

The aims of this population-based study were to elucidate the association between maternal overweight and obesity on the incidence of cerebral palsy, and to examine whether an association could be mediated by overweight or obesity–related pregnancy or neonatal complications.

The study was approved by the Research Ethics Committee at Karolinska Institutet, Stockholm, Sweden. The need for participants’ consent was waived because the data were anonymized. This population-based retrospective cohort study included live singleton births at 22 or more completed gestational weeks, recorded in the Swedish Medical Birth Register from 1997 through 2011. Using the person-unique national registration numbers of mothers and children,¹⁵ the Medical Birth Register was cross-linked with the National Patient, Cause of Death, Total Population, and Education Registers.

The Medical Birth Register includes data on more than 98% of all births in Sweden.¹⁶ Information on prenatal, obstetric, and neonatal care is prospectively recorded on standardized forms that are forwarded to the registry after the mother and infant are discharged from the hospital.¹⁶ The National Patient Register includes diagnostic information on hospital admissions since 1987 and hospital out-patient care from 2001.^17,18 The Cause of Death Register includes information on all deaths in Sweden.¹⁹ Diagnoses and causes of death are coded according to the Swedish version of the International Classification of Diseases, Tenth Revision (ICD-10) since 1997. The Total Population Register contains information on country of birth.²⁰ The Education Register records the highest level of education and is updated yearly.²¹

Maternal BMI (calculated as weight in kilograms divided by height in meters squared) in early pregnancy was calculated from self-reported height and weight measured in light clothing at the first prenatal visit, which occurs within the first 14 weeks of gestation for 90% of pregnant women.¹⁶ The correlation of height values between consecutive pregnancies by the same mother is 0.98.²² Hence, to reduce measurement error, we computed the median height across pregnancies for multiparous women.²³ Maternal BMI was classified according to the World Health Organization as underweight (<18.5), normal weight (18.5-24.9), overweight (25.0-29.9), obesity grade 1 (30.0-34.9), obesity grade 2 (35.0-39.9), or obesity grade 3 (≥40.0).²⁴ We also considered BMI as a continuous exposure.

Maternal age at delivery was calculated as date of delivery minus mother’s birth date. Parity was defined as the number of births of each mother. Information on whether the mother cohabited with the child’s father, a socioeconomic status indicator, was obtained at the first prenatal visit. Mothers were asked about smoking at the first prenatal visit and at 30 to 32 gestational weeks. Those who reported smoking at either visit were classified as smokers, whereas mothers who only stated that they were nonsmokers were classified as nonsmokers. This approach has been validated using cotinine markers.²⁵ Mothers’ country of birth and educational level were also considered.

All women in Sweden are offered ultrasound dating of pregnancy no later than early in the second trimester, and 95% accept.²⁶ Gestational age in completed weeks was estimated by early second trimester ultrasound dating in 87.7%, the date of the last menstrual period in 7.4%, and postnatal assessment in 4.9%. The length of gestation was categorized as full term (≥37 weeks), moderately preterm (32-36 weeks), very preterm (28-31 weeks), or extremely preterm (22-27 weeks). We examined the association of maternal BMI with cerebral palsy stratified by categories of gestational length.

We considered known consequences of maternal obesity as potential mediators of the association between early pregnancy BMI and cerebral palsy, including mode of delivery, traumatic delivery, macrosomia (birth weight ≥4000 g), neonatal infections and asphyxia-related neonatal morbidities (both within 28 days after birth), low Apgar score (<7) at 5 minutes, and congenital malformations. Categorization of covariates is provided in Table 1 and Table 2 and ICD-10 codes for diseases in mothers and children are provided in eTable 1 in the Supplement.

Information on cerebral palsy was based on diagnostic codes from the National Patient Register (ICD-10 code G80), diagnosed from 1997 through 2012. The quality of cerebral palsy diagnoses and registration in Sweden is high. About 40% of included cases are from country regions with well-established population-based cerebral palsy registries (Western Sweden since 1954 and South of Sweden since 1990). In these registries, all reported cases are systematically validated through medical record review, physical examination, or both by pediatric neurologists who are experts on cerebral palsy. Children from other regions are typically diagnosed by child neurologists, and a few by pediatricians experienced in child neurology. During the study period, children with cerebral palsy were entered in the Swedish National Quality Registry, which includes guidelines for diagnosis and classification.

The primary outcome was cerebral palsy of any type. We also considered the associations between early pregnancy BMI and subtypes of cerebral palsy (ICD-10 codes in eTable 1 in the Supplement).

Because the time of follow-up differed between participants, we estimated person-time incidence rates of cerebral palsy. Children were followed from birth to age at diagnosis of cerebral palsy, death, or end of follow-up (December 31, 2012), whichever came first. We first compared unadjusted rates between categories of maternal characteristics at baseline, pregnancy complications, and offspring diagnoses. Because the probability of being diagnosed with cerebral palsy changes with age, we calculated hazard ratios (HRs) and 95% CIs with the use of Cox proportional hazards models in which the outcome was age at first diagnosis of cerebral palsy. Next, we estimated rates of cerebral palsy in each BMI category. Adjusted HRs were obtained from multivariable Cox models with covariates including maternal age, country of origin, education level, cohabitation with partner, height, smoking, and year of delivery. For the computation of 95% CIs, we specified the robust sandwich estimate of the covariance matrix in all Cox models to account for the correlation of measures among multiparous women. Tests for linear trend were conducted by introducing a variable representing ordinal categories of BMI as a continuous predictor into the models. To assess potential nonlinear associations between BMI and cerebral palsy incidence, we estimated HRs for cerebral palsy in 13 fine BMI categories (eFigure 1A in the Supplement). In addition, we performed Cox regression with linear and nonlinear terms for BMI as a continuous predictor. Nonlinear terms were derived from restricted cubic spline models²⁷ with 3 to 9 knots (eFigure 1B in the Supplement). We used a stepwise algorithm to test the fit of the nonlinear (spline) terms. None of these terms offered a significantly better fit than the linear term for BMI alone in any of the models tested. Hence, we estimated the HR per unit BMI assuming that the association with cerebral palsy was linear (eFigure 1C in the Supplement). Because the validity of a cerebral palsy diagnosis may differ between young infants and older children, we conducted supplemental analyses after excluding infant deaths and infants diagnosed with cerebral palsy before 1 year of age.

Because maternal BMI is associated with risk of infant mortality²⁸ and asphyxia-related morbidity⁸ in infants born at full term, we decided a priori to estimate rates of cerebral palsy by BMI categories separately for full term, moderately preterm, very preterm, and extremely preterm infants. We tested the cross-product interaction terms between BMI and length of gestation categories with the use of a Wald χ² test. Maternal obesity is related to medically indicated preterm delivery through increased incidence of obstetric complications (eg, preeclampsia) and to spontaneous extremely preterm delivery.^7,22 In addition, preterm delivery is a strong risk factor for cerebral palsy. Therefore, we assessed whether an association between maternal BMI and cerebral palsy was mediated by preterm delivery using a counterfactual approach. We laid out a causal diagram²⁹ (Figure, A) to represent the potential mediating effect of preterm delivery in the presence of common causes (cause set 1) of exposure (BMI), mediator (preterm delivery), and outcome (cerebral palsy); and common causes (cause set 2) of mediator and outcome. We estimated the direct association of BMI on cerebral palsy independent of preterm delivery, the indirect association of BMI mediated through preterm delivery, and the proportion of the total BMI association mediated through preterm delivery using Valeri and VanderWeele’s formulas.³⁰ These formulas allow for exposure-mediator interactions and have been extended to the context of time-to-event analysis with Cox regression.³¹ The method assumes no unmeasured confounding (common causes) in the exposure-outcome, mediator-outcome, and exposure-mediator relations, and no effect of the exposure on confounders of the mediator-outcome relation (ie, no arrow from BMI to cause set 2 in Figure, A). Because the association of BMI with cerebral palsy was linear, for parsimony we used BMI as a continuous variable in mediation analyses. Also, because the proportion of the total effect mediated varies according to the exposure value at which it is estimated,³¹ we used as the exposure the difference between the median BMI of women with obesity (32.8) and those without obesity (23.1) , which was approximately 10 units. The mediator—preterm delivery—was defined dichotomously as less than 37 or 37 or more gestational weeks and the outcome as time to cerebral palsy diagnosis or censoring. The exposure-mediator and exposure-outcome relations were modeled with logistic and Cox regression, respectively, adjusted for sets of confounders (cause set 1: maternal age, country of origin, education level, cohabitation with a partner, parity, height, year of delivery; cause set 2:smoking during pregnancy and child’s sex) as categorical variables, and including an interaction term between BMI and preterm delivery.

Because most children with cerebral palsy are born at full term, we conducted further mediation analyses of the association between early pregnancy BMI and cerebral palsy in full-term children. Quiz Ref IDThe mediators considered were known consequences of maternal obesity that are also possible causes of cerebral palsy, including instrumental (cesarean or vaginal) delivery, traumatic delivery, macrosomia (birth weight >4000 g), neonatal infections, asphyxia-related neonatal complications (meconium aspiration, hypoxic ischemic encephalopathy, or neonatal seizures), Apgar score less than 7 at 5 minutes, presence of any congenital malformation, and presence of nervous system malformations. The causal diagram was modified (Figure, B) to represent the restriction on gestational age (box enclosing preterm delivery), each potential mediator, and a set of possible common causes (cause set 3: intrauterine growth) of the mediator and the outcome (cerebral palsy). To restrict to full-term children, we assumed that there were no unmeasured common causes of length of gestation and mediators or cerebral palsy.

All mediators were defined as dichotomous variables. One set of models for each potential mediator was fitted to estimate direct and indirect associations with BMI, following the same counterfactual approach described above. A BMI-mediator interaction term was included to estimate the association of BMI on cerebral palsy by levels of the mediator; this term was retained in the estimation of effects when it was statistically significant. In addition to the cause sets 1 and 2 , all but the models for congenital malformations also included birth weight for gestational age (cause set 3). All mediation analyses were conducted with use of the %mediation macro³¹ for SAS, version 9.4.

A sizeable proportion (10.5%) of women were excluded from the primary analysis due to missing data on BMI. Because exclusions due to missing data may lead to selection bias, we conducted supplemental analyses using multiple imputation of missing values. Missing values on BMI and covariates were estimated using a Markov Chain Monte Carlo multiple imputation method before their inclusion in the multivariable models.³² The variables entered in the imputation method were maternal age, country of origin, education level, cohabitation with a partner, parity, height, smoking during pregnancy, year of delivery, BMI, gestational age at delivery, and cerebral palsy diagnosis of the offspring. Results from 10 multiple imputation cycles were combined with use of the MIANALYZE procedure of SAS.

In additional supplemental analyses, we examined the associations of maternal BMI with cerebral palsy subtypes, as defined by ICD-10 codes (eTable 1 in the Supplement). A 2-tailed P value of <.05 was used as the threshold for statistical significance in all tests.

From 1997 through 2011, the Birth Register recorded information for 1 441 623 live singleton births. After excluding 17 694 births in which either the mothers or the children had invalid national registration numbers, 1 423 929 live singleton births (98.8%) were included in the study. Mean gestational age at delivery was 39.8 (SD, 1.8) weeks and 51.4% of the children were male. Compared with women with data on BMI, those excluded had higher rates of cerebral palsy and were slightly older and more likely to be of non-Nordic origin, had higher parity, were less likely to cohabit with a partner, were taller, and smoked more during pregnancy, and their children had higher rates of preterm delivery, low birth weight, neonatal infections, low Apgar score, and congenital malformations (eTable 2 in the Supplement).

There were 3029 children diagnosed with cerebral palsy over a median 7.8 years of follow-up (interquartile range, 4.3-11.7). The risk of cerebral palsy was 2.13 per 1000 live births and the incidence rate was 2.63 per 10 000 child-years. Rates of cerebral palsy were positively associated with low maternal education, non-Nordic origin, not cohabiting with a partner, primiparity or multiparity (≥4), short maternal stature, smoking during pregnancy, and maternal diabetic or hypertensive diseases (Table 1). Rates of cerebral palsy increased with male sex, decreasing length of gestation, instrumental or traumatic delivery, birth weight for-gestational age less than the 10th or more than the 97th percentile, neonatal infections, and congenital malformations (including chromosomal abnormalities and circulatory malformations) (Table 2). Malformations of the nervous system, asphyxia-related neonatal morbidities, and a low Apgar score (<7) at 5 minutes were associated with the largest increases in cerebral palsy rates (Table 2).

Quiz Ref IDMean early pregnancy BMI was 24.5 (SD, 4.4). The percentages of women in BMI categories were 2.4% for underweight (n = 30 765), 61.8% for normal weight (n = 787 815), 24.8% for overweight (n = 316 011), 7.8% for obesity grade 1 (n = 99 950), 2.4% for obesity grade 2 (n = 29 933), and 0.8% for obesity grade 3 (n = 10 413). Early pregnancy BMI was positively related to cerebral palsy rates. The number of cerebral palsy cases in each BMI category was 64, 1487, 728, 239, 88, and 38; and the rates per 10 000 child-years were 2.58, 2.35, 2.92, 3.15, 4.00, and 5.19, respectively. Compared with children of normal weight mothers, the adjusted HR of cerebral palsy for overweight was 1.22 (95% CI, 1.11-1.33) and 1.28 (95% CI, 1.11, 1.47) for obesity grade 1, 1.54 (95% CI, 1.24-1.93) for grade 2, and 2.02 (95% CI, 1.46-2.79) for grade 3 (Table 3). Results were essentially unchanged in supplemental analyses of diagnoses made after 1 year of age only (eTable 3 in the Supplement) or in analyses with multiple imputation of missing data (eTable 4 in the Supplement).

Of all children with cerebral palsy, 71% were born at full term, 13% moderately preterm, 10% very preterm, and 6% extremely preterm. Gestational length modified the association between maternal BMI and rates of cerebral palsy (P for interaction = .003). The rates of cerebral palsy increased with maternal overweight and obesity severity only in children born at full term (Table 4). In children born at full term, the numbers of cerebral palsy cases and rates per 10 000 child-years in maternal BMI categories were 1056 cases (rate, 1.75) for normal weight, 525 cases (rate, 2.22) for overweight, 165 cases (rate, 2.31) for obesity grade 1, 69 cases (rate, 3.36) for grade 2, and 23 cases (rate, 3.41) for grade 3. Compared with normal weight women, adjusted HR for overweight and obesity categories were 1.26 (95% CI, 1.14-1.41) for overweight, 1.30 (95% CI, 1.10-1.54) for obesity grade 1, 1.78 (95% CI, 1.38-2.29) for grade 2, and 1.86 (95% CI, 1.23-2.81) for grade 3. Among children born moderately preterm, only obesity grade 3 was associated with increased rates of cerebral palsy, whereas among those born very or extremely preterm there was generally no association between maternal BMI and cerebral palsy rates. Results did not change in analyses with multiple imputation of missing data (eTable 5 in the Supplement). When preterm birth (<37 weeks) was considered as a potential mediator (Figure, A), the HR of a 10-unit BMI increase from the baseline level (median BMI of women without obesity, 23.1) was 1.32 (95% CI, 1.22-1.43) for total association, 1.29 (95% CI, 1.19-1.40) for direct association, and 1.02 (95% CI, 1.02-1.03) for indirect association. Only 9.5% of the association of BMI with cerebral palsy was mediated through preterm delivery.

In mediation analyses for consequences of maternal obesity that may in turn cause cerebral palsy among children born at full term (Figure, B and Table 5), Quiz Ref IDan estimated 45% of the association between BMI and cerebral palsy was mediated through asphyxia-related neonatal complications, whereas 30% was mediated through low Apgar score. Instrumental delivery and nervous system malformations mediated 17% and 13% of the association, respectively. Other conditions each mediated less than 10%.

Among children born at full term, the strongest associations between maternal BMI and cerebral palsy were with dyskinetic cerebral palsy and spastic diplegic cerebral palsy (eTable 6 in the Supplement).

In this nationwide Swedish study, maternal overweight and increasing grades of obesity were associated with increasing rates of cerebral palsy. The association was restricted to children born at full term and was partly mediated through asphyxia-related neonatal complications.

More than 70% of cerebral palsy cases occur in children born at full term; thus, elucidating the etiology and identifying possible preventable factors for cerebral palsy in this group is of high clinical relevance. In this study, the association of maternal BMI with cerebral palsy in full-term children was partly mediated through asphyxia-related neonatal complications and, to a lesser extent, through low Apgar score and instrumental delivery, which are also associated with perinatal asphyxia.^33,34 The relative contribution of asphyxia to the etiology of cerebral palsy is a matter of debate. In a review of 23 studies,³⁵ the proportion of cerebral palsy cases with asphyxia varied from 3% to 56%, probably due to differences in operational definitions of asphyxia, cerebral palsy, or both; small sample sizes; and selection problems. The conditions commonly used to define asphyxia, including hypoxic ischemic encephalopathy, seizures, and low Apgar score, are not necessarily specific for fetal hypoxia, but could be the result of other pathologies including chorioamnionitis, stroke, or inflammation.³⁵ For most cases, cerebral palsy is thought to be caused by multiple, often interrelated factors that may obscure a specific effect of neonatal asphyxia. However, maternal overweight and obesity are associated with increased risks of emergency cesarean deliveries and hypoxic ischemic encephalopathy, seizures, meconium aspiration, and low Apgar scores at 5 and 10 minutes in full-term newborns.^8,36 The effects of maternal obesity on severe neonatal asphyxia may be partly explained by traumatic labor,³⁷ which often results from macrosomia.⁶ Another consequence of maternal obesity is fetal hyperinsulinemia, which may be related to chronic hypoxia even without diabetes.³⁸ Other mechanisms proposed to explain an effect of maternal obesity on neonatal asphyxia include lipotoxicity, placental inflammation and vasculopathy, and cord coiling.⁸ Evidence has demonstrated altered gene expression in full-term newborns of mothers with obesity, involving dysregulation of brain development, inflammatory and immune signaling, glucose and lipid homeostasis, and oxidative stress.³⁹

The BMI-related increase in cerebral palsy rates in full-term children was strongest for dyskinetic cerebral palsy, the subtype most often attributed to perinatal asphyxia in children born at full term and most strongly related to low Apgar scores.⁴⁰ Children with dyskinetic cerebral palsy exhibit neuroradiological alterations in the basal ganglia or the thalamus, brain structures that are particularly sensitive to asphyxia at full term.^40,41

The study has several strengths. The nationwide design provided information on approximately 1.5 million births linked to other nationwide registries, reducing the possibility of selection bias. Exposure information was prospectively collected, and weight was measured in early pregnancy, minimizing recall bias and measurement error. In addition to adjustment for confounding factors, the relative homogeneity of the Swedish population and the existence of a publicly funded, standardized health care system should limit residual confounding and provide high internal validity. The large sample size allowed exploration of associations between maternal BMI and subtypes of cerebral palsy, and possible pathways that could mediate an effect of maternal BMI on rates of cerebral palsy.

There are also some limitations. Ten percent of women had missing information on BMI or important covariates, which could lead to selection bias. The use of a new cerebral palsy subtype classification system in Europe⁴² and the adoption of a new global definition and classification⁴³ that are slightly different from the international version of ICD-10⁴⁴ may have introduced some misclassification of subtypes. Because this lack of specificity is independent of maternal BMI, the associations between maternal BMI and specific cerebral palsy subtypes could be attenuated. Statistical power was lower for the analyses of cerebral palsy subtypes than for the main analysis. Because the Swedish population is relatively homogeneous, the results may not be generalizable to more heterogeneous populations.

Although the effect of maternal obesity on cerebral palsy may seem small compared with other risk factors, the association is of public health relevance due to the large proportion of women with overweight or obesity worldwide. The number of women with a BMI of 35 or more globally doubled from approximately 50 to 100 million from 2000 through 2010.⁴⁴ In the United States, approximately half of all pregnant women have overweight or obesity at the first prenatal visit.⁴⁵ Considering the high prevalence of obesity and the continued rise of its most severe forms, the finding that maternal overweight and obesity are related to rates of cerebral palsy in a dose-response manner may have serious public health implications.⁴⁶

Among Swedish women with singleton children, maternal overweight and obesity were significantly associated with the rate of cerebral palsy. The association was limited to children born at full term and was partly mediated through asphyxia-related neonatal complications.

Corresponding Author: Eduardo Villamor, MD, DrPH, Department of Epidemiology, University of Michigan School of Public Health, 1420 Washington Heights, Ann Arbor, MI 48109 (villamor@umich.edu).

Author Contributions: Drs Villamor and Cnattingius had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Villamor, Tedroff, Peterson, Cnattingius.

Acquisition, analysis, or interpretation of data: Villamor, Tedroff, Johansson, Neovius, Petersson, Cnattingius.

Drafting of the manuscript: Villamor, Cnattingius.

Critical revision of the manuscript for important intellectual content: Tedroff, Peterson, Johansson, Neovius, Petersson, Cnattingius.

Statistical analysis: Villamor.

Obtained funding: Cnattingius.

Administrative, technical, or material support: Peterson, Neovius, Petersson, Cnattingius.

Supervision: Villamor, Tedroff, Cnattingius.

Other - Disease specific knowledge: Tedroff.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Neovius reports being a board member for Itrim and receiving grant funding from the National Institutes of Health for research comparing bariatric surgery and nonsurgical treatment for weight loss. No other disclosures were reported.

Funding/Support: The study was supported by grant 2014-0073 from the Swedish Research Council for Health, Working Life, and Welfare and an unrestricted grant from Karolinska Institutet (Dr Cnattingius).

Role of the Funder/Sponsor: The funding organizations for this study had no involvement in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication.