SummaryTwo approaches commonly used to deal with missing data are multiple imputation (MI) and inverse-probability weighting (IPW). IPW is also used to adjust for unequal sampling fractions. MI is generally more efficient than IPW but more complex. Whereas IPW requires only a model for the probability that an individual has complete data (a univariate outcome), MI needs a model for the joint distribution of the missing data (a multivariate outcome) given the observed data. Inadequacies in either model may lead to important bias if large amounts of data are missing. A third approach combines MI and IPW to give a doubly robust estimator. A fourth approach (IPW/MI) combines MI and IPW but, unlike doubly robust methods, imputes only isolated missing values and uses weights to account for remaining larger blocks of unimputed missing data, such as would arise, e.g., in a cohort study subject to sample attrition, and/or unequal sampling fractions. In this article, we examine the performance, in terms of bias and efficiency, of IPW/MI relative to MI and IPW alone and investigate whether the Rubin’s rules variance estimator is valid for IPW/MI. We prove that the Rubin’s rules variance estimator is valid for IPW/MI for linear regression with an imputed outcome, we present simulations supporting the use of this variance estimator in more general settings, and we demonstrate that IPW/MI can have advantages over alternatives. IPW/MI is applied to data from the National Child Development Study.

SummaryBackgroundSocioeconomic inequalities in childhood body-mass index (BMI) have been documented in high-income countries; however, uncertainty exists with regard to how they have changed over time, how inequalities in the composite parts (ie, weight and height) of BMI have changed, and whether inequalities differ in magnitude across the outcome distribution. Therefore, we aimed to investigate how socioeconomic inequalities in childhood and adolescent weight, height, and BMI have changed over time in Britain.MethodsWe used data from four British longitudinal, observational, birth cohort studies: the 1946 Medical Research Council National Survey of Health and Development (1946 NSHD), 1958 National Child Development Study (1958 NCDS), 1970 British Cohort Study (1970 BCS), and 2001 Millennium Cohort Study (2001 MCS). BMI (kg/m2) was derived in each study from measured weight and height. Childhood socioeconomic position was indicated by the father's occupational social class, measured at the ages of 10–11 years. We examined associations between childhood socioeconomic position and anthropometric outcomes at age 7 years, 11 years, and 15 years to assess socioeconomic inequalities in each cohort using gender-adjusted linear regression models. We also used multilevel models to examine whether these inequalities widened or narrowed from childhood to adolescence, and quantile regression was used to examine whether the magnitude of inequalities differed across the outcome distribution.FindingsIn England, Scotland, and Wales, 5362 singleton births were enrolled in 1946, 17 202 in 1958, 17 290 in 1970, and 16 404 in 2001. Low socioeconomic position was associated with lower weight at childhood and adolescent in the earlier-born cohorts (1946–70), but with higher weight in the 2001 MCS cohort. Weight disparities became larger from childhood to adolescence in the 2001 MCS but not the earlier-born cohorts (pinteraction=0·001). Low socioeconomic position was also associated with shorter height in all cohorts, yet the absolute magnitude of this difference narrowed across generations. These disparities widened with age in the 2001 MCS (pinteraction=0·002) but not in the earlier-born cohorts. There was little inequality in childhood BMI in the 1946–70 cohorts, whereas inequalities were present in the 2001 cohort and widened from childhood to adolescence in the 1958–2001 cohorts (pinteraction<0·05 in the later three cohorts but not the 1946 NSHD). BMI and weight disparities were larger in the 2001 cohort than in the earlier-born cohorts, and systematically larger at higher quantiles—eg, in the 2001 MCS at age 11 years, a difference of 0·98 kg/m2 (95% CI 0·63–1·33) in the 50th BMI percentile and 2·54 kg/m2 (1·85–3·22) difference at the 90th BMI percentile were observed.InterpretationOver the studied period (1953–2015), socioeconomic-associated inequalities in weight reversed and those in height narrowed, whereas differences in BMI and obesity emerged and widened. These substantial changes highlight the impact of societal changes o...

BackgroundThere is a paucity of information on secular trends in the age-related process by which people develop overweight or obesity. Utilizing longitudinal data in the United Kingdom birth cohort studies, we investigated shifts over the past nearly 70 years in the distribution of body mass index (BMI) and development of overweight or obesity across childhood and adulthood.Methods and FindingsThe sample comprised 56,632 participants with 273,843 BMI observations in the 1946 Medical Research Council National Survey of Health and Development (NSHD; ages 2–64 years), 1958 National Child Development Study (NCDS; 7–50), 1970 British Cohort Study (BCS; 10–42), 1991 Avon Longitudinal Study of Parents and Children (ALSPAC; 7–18), or 2001 Millennium Cohort Study (MCS; 3–11). Growth references showed a secular trend toward positive skewing of the BMI distribution at younger ages. During childhood, the 50th centiles for all studies lay in the middle of the International Obesity Task Force normal weight range, but during adulthood, the age when a 50th centile first entered the overweight range (i.e., 25–29.9 kg/m2) decreased across NSHD, NCDS, and BCS from 41 to 33 to 30 years in males and 48 to 44 to 41 years in females. Trajectories of overweight or obesity showed that more recently born cohorts developed greater probabilities of overweight or obesity at younger ages. Overweight or obesity became more probable in NCDS than NSHD in early adulthood, but more probable in BCS than NCDS and NSHD in adolescence, for example. By age 10 years, the estimated probabilities of overweight or obesity in cohorts born after the 1980s were 2–3 times greater than those born before the 1980s (e.g., 0.229 [95% CI 0.219–0.240] in MCS males; 0.071 [0.065–0.078] in NSHD males). It was not possible to (1) model separate trajectories for overweight and obesity, because there were few obesity cases at young ages in the earliest-born cohorts, or (2) consider ethnic minority groups. The end date for analyses was August 2014.ConclusionsOur results demonstrate how younger generations are likely to accumulate greater exposure to overweight or obesity throughout their lives and, thus, increased risk for chronic health conditions such as coronary heart disease and type 2 diabetes mellitus. In the absence of effective intervention, overweight and obesity will have severe public health consequences in decades to come.

No abstract

High BMI and excessive BMI gain at any life stage, particularly recently, is associated with increased adult BP. Relatively thin children were vulnerable to BMI gain and increased BP in adulthood. Elevated BP associated with low birthweight was not caused by 'catch-up' growth.