Cohort profile: 1958 British birth cohort (National Child Development Study)

Changes in the behavioural determinants of health during the COVID-19 pandemic: gender, socioeconomic and ethnic inequalities in five British cohort studies

BACKGROUND

The COVID-19 pandemic is expected to have far-reaching consequences on population health. We investigated whether these consequences included changes in health-impacting behaviours which are important drivers of health inequalities.

METHODS

Using data from five representative British cohorts (born 2000-2002, 1989-1990, 1970, 1958 and 1946), we investigated sleep, physical activity (exercise), diet and alcohol intake (N=14 297). We investigated change in each behaviour (pre/during the May 2020 lockdown), and differences by age/cohort, gender, ethnicity and socioeconomic position (childhood social class, education attainment and adult financial difficulties). Logistic regression models were used, accounting for study design and non-response weights, and meta-analysis used to pool and test cohort differences in association.

RESULTS

Change occurred in both directions-shifts from the middle part of the distribution to both declines and increases in sleep, exercise and alcohol use. Older cohorts were less likely to report changes in behaviours while the youngest reported more frequent increases in sleep, exercise, and fruit and vegetable intake, yet lower alcohol consumption. Widening inequalities in sleep during lockdown were more frequent among women, socioeconomically disadvantaged groups and ethnic minorities. For other outcomes, inequalities were largely unchanged, yet ethnic minorities were at higher risk of undertaking less exercise and consuming lower amounts of fruit and vegetables.

CONCLUSIONS

Our findings provide new evidence on the multiple changes to behavioural outcomes linked to lockdown, and the differential impacts across generation, gender, socioeconomic circumstances across life, and ethnicity. Lockdown appeared to widen some (but not all) forms of health inequality.

Depressive symptoms during early adulthood and the development of physical multimorbidity in the UK: an observational cohort study

BACKGROUND

An understanding of whether early-life depression is associated with physical multimorbidity could be instrumental for the development of preventive measures and the integrated management of depression. We therefore aimed to map out the cumulative incidence of physical multimorbidity over adulthood, and to determine the association between the presence of depressive symptoms during early adulthood and the development of physical multimorbidity in middle age.

METHODS

In this observational cohort study, we used pooled data from the 1958 National Child Development Study (NCDS) and the 1970 British Cohort Study (BCS). Cohort waves were pooled in each decade of adult life available (when cohort members were aged 26 years in the BCS and 23 years in the NCDS [baseline]; 34 years in the BCS and 33 years in the NCDS [age 34 BCS/33 NCDS]; 42 years in the BCS and NCDS [age 42 BCS/NCDS]; and 46 years in the BCS and 50 years in the NCDS [age 46 BCS/50 NCDS]). We included participants who had completed the nine-item Malaise Inventory at baseline, and did not have a history of physical multimorbidity, any physical multimorbidity at baseline, or the presence of depressive symptoms before the development of physical multimorbidity. The presence of depressive symptoms was determined using the nine-item Malaise Inventory (cutoff score ≥4). Physical multimorbidity was defined as having at least two measures of any of the following ten self-reported groups of long-term conditions: asthma or bronchitis; backache; bladder or kidney conditions; cancer; cardiovascular conditions; convulsions or epilepsy; diabetes; hearing conditions; migraine; and stomach, bowel, or gall conditions. Cumulative incidence (with 95% CI) of physical multimorbidity was calculated for each decade considered after baseline, with physical multimorbidity being assessed as both a dichotomous and categorical variable. The association between depressive symptoms and the development of physical multimorbidity was assessed using adjusted relative risk ratios (with 95% CIs).

FINDINGS

Analyses included 15 845 participants, of whom 4001 (25·25%; 95% CI 24·57-25·93) had depressive symptoms at baseline and 11 844 (74·75%; 74·07-75·42) did not. The cumulative incidence of physical multimorbidity (dichotomous) ranged over the study period from 2263 (18·44%; 95% CI 17·75-18·14) of 12 273 participants at age 34 BCS/33 NCDS, to 4496 (42·90%; 41·95-43·85) of 10 481 participants at age 46 BCS/50 NCDS, and was consistently higher in participants with depressive symptoms at baseline. The adjusted relative risk of physical multimorbidity was higher in participants with depressive symptoms than in those without and remained stable over the study period (adjusted relative rate ratio 1·67, 95% CI 1·50-1·87, at age 34 BCS/33 NCDS; 1·63, 1·48-1·79, at age 42 BCS/NCDS; and 1·58, 1·43-1·73, at age 46 BCS/50 NCDS).

INTERPRETATION

The presence of depressive symptoms during early adulthood is associated with an increased risk of the development of physical multimorbidity in middle age. Although further research about the drivers of this relationship is needed, these results could help to enhance the integrated management of individuals with depressive symptoms and the development of preventive strategies to reduce the effect and burden of physical multimorbidity.

FUNDING

UK Medical Research Council and Guy's Charity.

Changes in the body mass index and blood pressure association across time: Evidence from multiple cross-sectional and cohort studies

Although body mass index (BMI) is considered a key determinant of high blood pressure, its importance may differ over time and by age group. We utilised separate data sources to investigate temporal changes in this association: 23 independent (newly sampled), repeated cross-sectional studies (Health Survey for England (HSE)) at ≥25 years (1994-2018; N = 126,742); and three British birth cohorts at 43-46 years (born 1946, 1958, and 1970; N = 18,657). In HSE, associations were weaker in more recent years, with this trend most pronounced amongst older adults. After adjustment for sex, anti-hypertensive treatment and education, the mean difference in systolic blood pressure (SBP) per 1 kg/m² increase in BMI amongst adults ≥55 years was 0.75 mmHg (95%CI: 0.60-0.90) in 1994, 0.66 mmHg (0.46-0.85) in 2003, and 0.53 mmHg (0.35-0.71) in 2018. In the 1958 and 1970 cohorts, BMI and SBP associations were of similar magnitude yet weaker in the 1946 cohort, potentially due to differences in blood pressure measurement device. Quantile regression analyses suggested that associations between BMI and SBP were present both below and above the hypertension threshold. A weaker association between BMI and blood pressure may partly offset the public health impacts of increasing obesity prevalence. However, despite sizable increases in use of antihypertensive medication, BMI remains positively associated with SBP in all ages. Our findings highlight the need to tackle non-medical factors such as population diet which influence both BMI and blood pressure, and the utility of using multiple datasets to obtain robust inferences on trends in risk factor-outcome associations across time.

Adult retrospective report of child abuse and prospective indicators of childhood harm: a population birth cohort study

BACKGROUND

We aim to determine whether adult retrospective report of child abuse is associated with greater risk of prospectively assessed harmful environments in childhood. We assessed possible recall basis by adult depression status.

METHODS

At 45 years, participants of the 1958 British birth cohort (N = 9308) reported a range of abuse types (by 16 years). Prospective data, ages 7-16 years, were obtained for impoverished upbringing, hazardous conditions, anti-social behaviours and 16 years poor parent-child relationships. We estimated associations between retrospective report of child abuse and prospectively measured harm using (i) odds ratios (ORs, 95% confidence intervals) and (ii) positive predictive values (PPVs). PPVs were calculated stratified by adult depression status.

RESULTS

Prevalence of retrospectively reported abuse ranged from 10.7% (psychological) to 1.60% (sexual) and 14.8% reported ≥ 1 type; prospectively recorded harm ranged from 10% (hazardous conditions/poor parent-child relationships) to 20% (anti-social behaviours). Adults retrospectively reporting abuse were more likely to have had harmful childhood environments: 52.4% had ≥ 1 indicator of harm (vs. 35.6% among others); OR_sex-adjusted for poor relationships with parents was 2.98 (2.50, 3.54). For retrospectively reported (vs. none) abuse, there was a trend of increasing relative risk ratio with number of harms, from 1.75 (1.50, 2.03) for 1 to 4.68 (3.39, 6.45) for 3/4 childhood harms. The PPV of ≥ 1 prospectively recorded harm did not differ between depressed (0.58 (0.52, 0.64)) and non-depressed (0.58 (0.55, 0.61)) groups.

CONCLUSIONS

In a population cohort, adult retrospective report of child abuse was associated with several harms, prospectively measured from childhood to adolescence, providing support for the validity of retrospective report-based research. Findings suggest retrospectively reported child abuse is not biased by depression in adulthood.

Irritable bowel syndrome and risk of glaucoma: An analysis of two independent population-based cohort studies

OBJECTIVE

Irritable bowel syndrome (IBS) is a chronic disorder associated with an abnormal gastrointestinal microbiome. Microbiome-host interactions are known to influence organ function including in the central nervous system; thus, we sought to identify whether IBS may be a risk factor for the development of glaucoma.

DESIGN

Two prospective cohort studies.

SUBJECTS

The 1958 United Kingdom Birth Cohort (UKBC; 9091 individuals) and the Danish National Registry of Patients (DNRP; 62,541 individuals with IBS and 625,410 matched general population cohort members).

METHODS

In the UKBC, participants were surveyed throughout life (including at ages 42 and 50). The DNRP contains records of hospital-based contacts and prescription data from the national prescription database.

MAIN OUTCOME MEASURE

The main outcome measure was incidence of glaucoma. In the UKBC, incident glaucoma at age 50 (n = 48) was determined through comparison of survey responses at ages 42 and 50 years. In the DNRP, glaucoma was assessed by hospital diagnosis (n = 1510), glaucoma surgery (n = 582) and initiation of glaucoma medications (n = 1674).

RESULTS

In the UKBC, the odds ratio (OR) of developing glaucoma between ages 42 and 50 in persons with a chronic IBS diagnosis was increased [OR: 5.84, 95% confidence interval (CI): 2.26-15.13]. People with an IBS diagnosis in the DNRP had a hazard ratio (HR) of 1.35 for developing physician-diagnosed glaucoma (95% CI: 1.16-1.56), an HR of 1.35 for undergoing glaucoma surgery (95% CI: 1.06-1.70) and an HR of 1.19 for initiating glaucoma medication (95% CI: 1.03-1.38).

CONCLUSIONS

In two large European cohort studies, IBS is a risk factor for glaucoma.

Genome-wide association study identifies susceptibility loci for acute myeloid leukemia

Acute myeloid leukemia (AML) is a hematological malignancy with an undefined heritable risk. Here we perform a meta-analysis of three genome-wide association studies, with replication in a fourth study, incorporating a total of 4018 AML cases and 10488 controls. We identify a genome-wide significant risk locus for AML at 11q13.2 (rs4930561; P = 2.15 × 10-8; KMT5B). We also identify a genome-wide significant risk locus for the cytogenetically normal AML sub-group (N = 1287) at 6p21.32 (rs3916765; P = 1.51 × 10-10; HLA). Our results inform on AML etiology and identify putative functional genes operating in histone methylation (KMT5B) and immune function (HLA).

Impact of lockdown on key workers: findings from the COVID-19 survey in four UK national longitudinal studies

BACKGROUND

Key workers played a pivotal role during the national lockdown in the UK's response to the COVID-19 pandemic. Although protective measures have been taken, the impact of the pandemic on key workers is yet to be fully elucidated.

METHODS

Participants were from four longitudinal age-homogeneous British cohorts (born in 2001, 1990, 1970 and 1958). A web-based survey provided outcome data during the first UK national lockdown (May 2020) on COVID-19 infection status, changes in financial situation, trust in government, conflict with people around, household composition, psychological distress, alcohol consumption, smoking and sleep duration. Generalised linear models with logit link assessed the association between being a key worker and the above outcomes. Adjustment was made for cohort design, non-response, sex, ethnicity, adult socioeconomic position (SEP), childhood SEP, the presence of a chronic illness and receipt of a shielding letter. Meta-analyses were performed across the cohorts.

FINDINGS

13 736 participants were included. During lockdown, being a key worker was associated with increased chances of being infected with COVID-19 (OR 1.43, 95% CI 1.22 to 1.68) and experiencing conflict with people around (OR 1.19, 95% CI 1.03 to 1.37). However, key workers were less likely to be worse off financially (OR 0.32, 95% CI 0.24 to 0.65), to consume more alcohol (OR 0.88, 95% CI 0.79 to 0.98) or to smoke more (OR 0.60, 95% CI 0.44 to 0.80) during lockdown. Interestingly, being a key worker was not associated with psychological distress (OR 0.95, 95% CI 0.85 to 1.05).

INTERPRETATION

Being a key worker during the first UK COVID-19 lockdown was a double-edged sword, with both benefits and downsides. The UK government had the basic duty to protect its key workers from SARS-CoV-2 infection, but it may have failed to do so, and there is an urgent need to rectify this in light of the ongoing third wave.

Patterns of Atopic Eczema Disease Activity From Birth Through Midlife in 2 British Birth Cohorts

Importance

Atopic eczema is characterized by a heterogenous waxing and waning course, with variable age of onset and persistence of symptoms. Distinct patterns of disease activity such as early-onset/resolving and persistent disease have been identified throughout childhood; little is known about patterns into adulthood.

Objective

This study aimed to identify subtypes of atopic eczema based on patterns of disease activity through mid-adulthood, to examine whether early life risk factors and participant characteristics are associated with these subtypes, and to determine whether subtypes are associated with other atopic diseases and general health in mid-adulthood.

Design, Setting, and Participants

This study evaluated members of 2 population-based birth cohorts, the 1958 National Childhood Development Study (NCDS) and the 1970 British Cohort Study (BCS70). Participant data were collected over the period between 1958 and 2016. Data were analyzed over the period between 2018 and 2020.

Main Outcomes and Measures

Subtypes of atopic eczema were identified based on self-reported atopic eczema period prevalence at multiple occasions. These subtypes were the outcome in models of early life characteristics and an exposure variable in models of midlife health.

Results

Latent class analysis identified 4 subtypes of atopic eczema with distinct patterns of disease activity among 15 939 individuals from the NCDS (51.4% male, 75.4% White) and 14 966 individuals from the BCS70 (51.6% male, 78.8% White): rare/no (88% to 91%), decreasing (4%), increasing (2% to 6%), and persistently high (2% to 3%) probability of reporting prevalent atopic eczema with age. Sex at birth and early life factors, including social class, region of residence, tobacco smoke exposure, and breastfeeding, predicted differences between the 3 atopic eczema subtypes and the infrequent/no atopic eczema group, but only female sex differentiated the high and decreasing probability subtypes (odds ratio [OR], 1.99; 95% CI, 1.66-2.38). Individuals in the high subtype were most likely to experience asthma and rhinitis, and those in the increasing subtype were at higher risk of poor self-reported general (OR, 1.29; 95% CI, 1.09-1.53) and mental (OR 1.45; 95% CI, 1.23-1.72) health in midlife.

Conclusions and Relevance

The findings of this cohort study suggest that extending the window of observation beyond childhood may reveal clear subtypes of atopic eczema based on patterns of disease activity. A newly identified subtype with increasing probability of activity in adulthood warrants additional attention given observed associations with poor self-reported health in midlife.

Early life mental health and problematic drinking in mid-adulthood: evidence from two British birth cohorts

PURPOSE

Accumulating evidence suggests that externalising problems are consistently associated with alcohol use behaviours, but findings are inconsistent regarding the role of internalising problems. We investigate whether externalising and internalising problems are associated with problematic drinking in mid-adulthood, and whether potential associations are modified by age, sex and cohort.

METHODS

The National Child Development Study (NCDS58, n = 17,633) and 1970 British Cohort Study (BCS70, n = 17,568) recruited new-borns in Great Britain in a single week in 1958 and 1970. Mental health was assessed with the Rutter Behaviour Questionnaire at ages 7, 11, and 16 in NCDS58 and ages 5, 10 and 16 in BCS70. Problematic drinking was measured with the CAGE questionnaire at age 33 in NCDS58 and age 34 in BCS70, and the AUDIT scale at age 44/45 in NCDS58 and age 46 in BCS70. Latent scores of externalising and internalising problems were added chronologically into lagged logistic regression models.

RESULTS

Externalising and internalising problems were associated in opposite directions with problematic drinking in mid-adulthood. Externalising was a risk factor (OR [95% CI] ranging from 1.06 [1.03, 1.10] to 1.11 [1.07, 1.15] for different ages), and internalising was a protective factor (OR [95% CI] ranging from 0.95 [0.92, 0.99] to 0.90 [0.86, 0.94] for different ages). Associations between early life mental health and mid-adulthood problematic drinking did not differ by developmental timing but were stronger in males.

CONCLUSION

Our study provides new insights on links of externalising and internalising difficulties with alcohol use and has implications for public policy in the UK.

Rare and low-frequency exonic variants and gene-by-smoking interactions in pulmonary function

Genome-wide association studies have identified numerous common genetic variants associated with spirometric measures of pulmonary function, including forced expiratory volume in one second (FEV₁), forced vital capacity, and their ratio. However, variants with lower minor allele frequencies are less explored. We conducted a large-scale gene-smoking interaction meta-analysis on exonic rare and low-frequency variants involving 44,429 individuals of European ancestry in the discovery stage and sought replication in the UK BiLEVE study with 45,133 European ancestry samples and UK Biobank study with 59,478 samples. We leveraged data on cigarette smoking, the major environmental risk factor for reduced lung function, by testing gene-by-smoking interaction effects only and simultaneously testing the genetic main effects and interaction effects. The most statistically significant signal that replicated was a previously reported low-frequency signal in GPR126, distinct from common variant associations in this gene. Although only nominal replication was obtained for a top rare variant signal rs142935352 in one of the two studies, interaction and joint tests for current smoking and PDE3B were significantly associated with FEV₁. This study investigates the utility of assessing gene-by-smoking interactions and underscores their effects on potential pulmonary function.

Child maltreatment, early life socioeconomic disadvantage and all-cause mortality in mid-adulthood: findings from a prospective British birth cohort

OBJECTIVES

Early-life adversities (ELAs) such as child maltreatment (neglect and abuse) and socioeconomic disadvantage have been associated with adult mortality. However, evidence is sparse for specific types of ELA. We aimed to establish whether specific ELAs (ie, different types of child maltreatment and socioeconomic disadvantage) were associated independently with all-cause mortality in mid-adulthood and to examine potential intermediary pathways.

DESIGN

Prospective cohort study.

SETTING

1958 British birth cohort: a longitudinal, population-based sample of individuals born in Great Britain during a single week in March 1958.

PARTICIPANTS

9310 males and females with data on child maltreatment and mortality (44/45-58 years).

OUTCOME MEASURES

Mortality follow-up from 2002/2003 to 2016 when participants were aged 44/45-58 years. Death was ascertained via the NHS Central Register (N=296) or cohort maintenance activities (N=16).

RESULTS

Prevalence of ELAs ranged from 1.6% (sexual abuse) to 11% (psychological abuse). Several, but not all, ELAs were associated with increased risk of premature death, independent of covariates and other adversities; adjusted HRs were 2.64 (95% CI 1.52 to 4.59) for sexual abuse, 1.93 (95% CI 1.45 to 2.58) for socioeconomic disadvantage, 1.73 (95% CI 1.11 to 2.71) for physical abuse and 1.43 (95% CI 1.03 to 1.98) for neglect. After adjustment for covariates and other adversities, no associations with mortality were observed for psychological and witnessing abuse. Regarding potential intermediaries (including adult socioeconomic factors, behaviours, adiposity, mental health and cardiometabolic markers), most associations attenuated after accounting for adult health behaviours (particularly smoking). In addition, early-life socioeconomic disadvantage and neglect associations attenuated after accounting for adult socioeconomic factors. The association for sexual abuse and premature mortality was largely unaffected by potential intermediaries.

CONCLUSIONS

Associations with premature mortality varied by type of ELA: associations for sexual and physical abuse, neglect and socioeconomic disadvantage were independent of each other. Different types of ELAs could influence premature mortality via different pathways; this requires further research.

Does comprehensive education reduce health inequalities?

This article analyses the impact of comprehensive education on health inequalities. Given that education is an important social determinant of health, it is hypothesised that a more equitable comprehensive system could reduce health inequalities in adulthood. To test this hypothesis, we exploited the change from a largely selective to a largely comprehensive system that occurred in the UK from the mid-1960s onwards and compare inequalities in health outcomes of two birth cohorts (1958 and 1970) who attended either system. We studied physical and mental health, health behaviours and life satisfaction in middle age as outcomes and absolute and relative inequalities by social class (of origin and destination) and education. Inverse probability weighting was used to control confounding by socio-economic and education background, and ability test score taken prior to secondary school entry. We did not find consistent evidence that health inequalities were smaller under the comprehensive compared to the selective system and the results were robust under different model specifications. Our study adds to the sparse but growing literature that assesses the impact of social policy on health inequalities.

Clinical symptoms of anxiety disorders as predictors of political attitudes: A prospective cohort study

Conservative political ideologies have been suggested to correlate with elevated sensitivity to threat. However, it is unclear whether the associations between threat sensitivity and political attitudes can be observed with clinical measures of mental health. We examined how anxiety disorders predicted attitudes on several political issues. Participants were 7253 individuals from the 1958 British Birth Cohort study. Symptoms of generalised anxiety disorder, phobia and panic were assessed in a clinical interview at age 44, and opinions about political issues were self-reported by the participants 6 years later. Anxiety symptoms were associated with higher concerns about economic inequality, preservation of the environment, distrust in politics and lower work ethic. No associations were observed with racist or authoritarian attitudes, or support for traditional family values. We also assessed how political attitudes at ages 33 and 42 predicted anxiety disorder symptoms at age 44, revealing a possible bidirectional association between concern for economic inequality and anxiety disorder symptoms. These findings do not support an association between conservative political attitudes and elevated threat sensitivity. Rather, elevated anxiety may increase concerns about social inequality and the environment.

Childhood growth and development and DNA methylation age in mid-life

BACKGROUND

In the first study of its kind, we examine the association between growth and development in early life and DNAm age biomarkers in mid-life.

METHODS

Participants were from the Medical Research Council National Survey of Health and Development (n = 1376). Four DNAm age acceleration (AgeAccel) biomarkers were measured when participants were aged 53 years: AgeAccelHannum; AgeAccelHorvath; AgeAccelLevine; and AgeAccelGrim. Exposure variables included: relative weight gain (standardised residuals from models of current weight z-score on current height, and previous weight and height z-scores); and linear growth (standardised residuals from models of current height z-score on previous height and weight z-scores) during infancy (0-2 years, weight gain only), early childhood (2-4 years), middle childhood (4-7 years) and late childhood to adolescence (7-15 years); age at menarche; and pubertal stage for men at 14-15 years. The relationship between relative weight gain and linear growth and AgeAccel was investigated using conditional growth models. We replicated analyses from the late childhood to adolescence period and pubertal timing among 240 participants from The National Child and Development Study (NCDS).

RESULTS

A 1SD increase in relative weight gain in late childhood to adolescence was associated with 0.50 years (95% CI 0.20, 0.79) higher AgeAccelGrim. Although the CI includes the null, the estimate was similar in NCDS [0.57 years (95% CI - 0.01, 1.16)] There was no strong evidence that relative weight gain and linear growth in childhood was associated with any other AgeAccel biomarker. There was no relationship between pubertal timing in men and AgeAccel biomarkers. Women who reached menarche ≥ 12 years had 1.20 years (95% CI 0.15, 2.24) higher AgeAccelGrim on average than women who reached menarche < 12 years; however, this was not replicated in NCDS and was not statistically significant after Bonferroni correction.

CONCLUSIONS

Our findings generally do not support an association between growth and AgeAccel biomarkers in mid-life. However, we found rapid weight gain during pubertal development, previously related to higher cardiovascular disease risk, to be associated with older AgeAccelGrim. Given this is an exploratory study, this finding requires replication.

Early-Life Socioeconomic Position and the Accumulation of Health-Related Deficits by Midlife in the 1958 British Birth Cohort Study

Reducing population levels of frailty is an important goal, and preventing its development in midadulthood could be pivotal. There is limited evidence on associations between childhood socioeconomic position (SEP) and frailty. Using data on the 1958 British birth cohort (followed from 1958 to 2016; n = 8,711), we aimed to 1) establish the utility of measuring frailty in midlife, by examining associations between a 34-item frailty index at age 50 years (FI_50y) and mortality at ages 50–58 years, and 2) examine associations between early-life SEP and FI_50y and investigate whether these associations were explained by adult SEP. Hazard ratios for mortality increased with increasing frailty; for example, the sex-adjusted hazard ratio for the highest quintile of FI_50y versus the lowest was 4.07 (95% confidence interval (CI): 2.64, 6.25). Lower early-life SEP was associated with higher FI_50y. Compared with participants born in the highest social class, the estimated total effect on FI_50y was 42.0% (95% CI: 35.5, 48.4) for participants born in the lowest class, with the proportion mediated by adult SEP being 0.45% (95% CI: 0.35, 0.55). Mediation by adult SEP was negligible for other early-life SEP classes. Findings suggest that early-life SEP is associated with frailty and that adult SEP only partially explains this association. Results highlight the importance of improving socioeconomic circumstances across the life course to reduce inequalities in midlife frailty.

Comparison of smoking reduction with improvement of social conditions in early life: simulation in a British cohort

BACKGROUND

Health care evaluation models can be useful to assign different levels of priority to interventions or policies targeting different age groups or different determinants of health. We aimed to assess early mortality in counterfactual scenarios implying reduced adverse childhood experience (ACE) and/or improved educational attainment (childhood and early life characteristics), compared with a counterfactual scenario implying reduced smoking in adulthood.

METHODS

We used data from the 1958 National Child Development Study British birth cohort, which initially included 18 558 subjects. Applying a potential outcome approach, scenarios were simulated to estimate the expected mortality between ages 16 and 55 under a counterfactual decrease by half of the observed level of exposure to (i) ACE, (ii) low educational attainment (at age 22), (iii) ACE and low educational attainment (a combined exposure) and (iv) smoking at age 33. Estimations were obtained using g-computation, separately for men and women. Analyses were further stratified according to the parental level of education, to assess social inequalities.

RESULTS

The study population included 12 164 members. The estimated decrease in mortality in the counterfactual scenarios with reduced ACE and improved educational attainment was close to the decreased mortality in the counterfactual scenario with reduced smoking, showing a relative difference in mortality of respectively -7.2% [95% CI (confidence interval) = (-12.2% to 1.2%)] versus -7.0% (-13.1% to +1.2%) for women, and -9.9% (-15.6% to -6.2%) versus -12.3% (-17.0% to -5.9%) for men.

CONCLUSIONS

Our results highlight the potential value of targeting early social characteristics such as ACE and education, compared with well-recognized interventions on smoking.

Changes in emotional problems, hyperactivity and conduct problems in moderate to late preterm children and adolescents born between 1958 and 2002 in the United Kingdom

Background

Preterm birth is a risk factor for the development of emotional and behavioural problems in childhood and adolescence. Given the substantial improvements in neonatal care across decades, it has been expected that the difference in emotional problems, hyperactivity, and conduct problems between moderate to late preterm (MLPT) and full term (FT) children and adolescents have declined in recent years.

Methods

Data from four UK population-based studies were used: The National Child Development Study (NCDS; 1958), the British Cohort Study (BCS70; 1970), the Avon Longitudinal Study of Parents and Children (ALSPAC, 1991-1992) and the Millennium Cohort Study (MCS; 2000-2002). Emotional problems, hyperactivity and conduct problems were assessed with mother-reports in early childhood (5-7 years), late childhood (10-11 years) and adolescence (14-16 years). Furthermore, emotional problems were self-reported in adolescence in BCS70, ALSPAC and MCS.

Results

In the most recent cohort, the MLPT group had higher mother-reported emotional problems than those born FT in late childhood and adolescence, whereas there were no differences in self-reported emotional problems in adolescence. Regarding mother-reported hyperactivity symptoms, the MLPT group had higher scores than the FT group in the two most recent cohorts in late childhood and in the most recent cohort in adolescence. Regarding mother-reported conduct problems, MLPT children had significantly higher scores than the FT children in the oldest and in the most recent cohort in late childhood. On the other hand, in adolescence, MLPT children had significantly lower scores in conduct problems in comparison to FT children in the cohort born in 1991-1992.

Conclusions

Mother-reported emotional problems and hyperactivity symptoms of those born MLPT have increased from the birth years 1958 to 2000-2002 during late childhood and adolescence, whereas self-reported emotional problems were similar in MLPT and FT groups during adolescence from 1970 to 2000-2002. Findings are less consistent regarding conduct problems. The current findings highlight the importance of raising the awareness of teachers about the association between MLPT birth and behavioural and emotional problems in late childhood and adolescence to prevent the long-term negative outcomes associated with the sequalae of MLPT birth.

Influence of childhood socioeconomic position and ability on mid-life cognitive function: evidence from three British birth cohorts

BACKGROUND

Childhood socioeconomic position (SEP) is robustly associated with cognitive function later in life. However, it is unclear whether this reflects a direct relationship, or an indirect association via modifiable factors such as educational attainment and occupation. We sought to clarify these associations using retrospectively harmonised data from three ongoing British birth cohorts.

METHODS

We analysed data from the 1946 National Survey of Health and Development (n=2283), the 1958 National Child Development Study (n=9385) and the 1970 British Cohort Study (n=7631). Retrospective harmonisation was used to derive equivalent indicators of cognition, SEP, education and occupation across the three cohorts. Structural equation modelling was used to examine the association between childhood SEP and mid-life cognitive function, via childhood cognitive ability, educational attainment and mid-life occupation.

RESULTS

Across all three cohorts, no direct pathways were observed between childhood SEP and mid-life cognitive function. Rather, this association was indirect via the three temporally ordered mediators. In addition, the direct pathway between childhood cognition and adult cognitive function was weaker in the two younger studies.

CONCLUSIONS

Across three British birth cohorts, we found that the association between early life SEP and mid-life cognitive function was fully mediated by childhood cognitive ability, educational attainment and occupational status. Furthermore, the association between early cognitive ability and mid-life cognitive function has decreased in younger generations. Therefore, cognitive function in adulthood may be influenced by modifiable factors and societal change.

Socioeconomic inequalities in co-morbidity of overweight, obesity and mental ill-health from adolescence to mid-adulthood in two national birth cohort studies

AIM

To examine socioeconomic inequalities in comorbidity risk for overweight (including obesity) and mental ill-health in two national cohorts. We investigated independent effects of childhood and adulthood socioeconomic disadvantage on comorbidity from childhood to mid-adulthood, and differences by sex and cohort.

METHODS

Data were from 1958 National Child Development Study (NCDS58) and 1970 British Cohort Study (BCS70) [total N=30,868, 51% males] assessed at ages 10, 16, 23/26, 34 and 42 years. Socioeconomic indicators included childhood and adulthood social class and educational level. Risk for i. having healthy BMI and mental ill-health, ii. overweight and good mental health, and iii. overweight and mental ill-health was analysed using multinomial logistic regression.

FINDINGS

Socioeconomic disadvantage was consistently associated with greater risk for overweight-mental ill-health comorbidity at all ages (RRR 1.43, 2.04, 2.38, 1.64 and 1.71 at ages 10, 16, 23, 34 and 42 respectively for unskilled/skilled vs. professional/managerial class). The observed inequalities in co-morbidity were greater than those observed for either condition alone (overweight; RRR 1.39 and 1.25, mental ill-health; 1.36 and 1.22 at ages 16 and 42 respectively, for unskilled/skilled vs. professional/managerial class). In adulthood, both childhood and adulthood socioeconomic disadvantage were independently associated with comorbid overweight-mental ill-health, with a clear inverse gradient between educational level and risk for comorbidity. For instance, for the no education group (compared to university education) the RRR is 6.11 (95% CI 4.31-8.65) at age 34 and 4.42 (3.28-5.96) at age 42. There were no differences observed in the extent of inequalities by sex and differences between cohorts were limited.

INTERPRETATION

While socioeconomic disadvantage in childhood and adulthood are consistently and independently associated with greater risk for mental ill-health and being overweight separately, these associations are even larger for their comorbidity across the lifecourse. These findings are significant given the increasing global prevalence of obesity and mental ill-health, and their implications for lifelong health and mortality.

FUNDING

This research was supported by grants from the Wellcome Trust (ISSF3/ H17RCO/NG1) and Medical Research Council (MRC) [MC_UU_00019/3].

Could teacher-perceived parental interest be an important factor in understanding how education relates to later physiological health? A life course approach

Education is associated with later health, and notably with an indicator of physiological health measuring the cost of adapting to stressful conditions, named allostatic load. Education is itself the result of a number of upstream variables. We examined the origins of educational attainment through the lens of interactions between families and school i.e. parents’ interest in their child’s education as perceived by teachers. This study aims to examine whether parental interest during a child’s educational trajectory is associated with subsequent allostatic load, and whether education or other pathways mediate this relationship. We used data from 9 377 women and men born in 1958 in Great Britain and included in the National Child Development Study to conduct secondary data analyses. Parental interest was measured from questionnaire responses by teachers collected at age 7, 11 and 16. Allostatic load was defined using 14 biomarkers assayed in blood from a biosample collected at 44 years of age. Linear regression analyses were carried out on a sample of 8 113 participants with complete data for allostatic load, missing data were imputed. Participants whose parents were considered to be uninterested in their education by their teacher had a higher allostatic load on average in midlife in both men (β = 0,41 [0,29; 0,54]) and women (β = 0,69 [0,54; 0,83]). We examined the role of the educational and other pathways including psychosocial, material/financial, and behavioral variables, as potential mediators in the relationship between parental interest and allostatic load. The direct link between parental interest and allostatic load was completely mediated in men, but only partially mediated in women. This work provides evidence that parents’ interest in their child’s education as perceived by teachers is associated with subsequent physiological health in mid-life and may highlight a form of cultural dissonance between family and educational spheres.

Framingham Heart Study: JACC Focus Seminar, 1/8

The Framingham Heart Study is the longest-running cardiovascular epidemiological study, starting in 1948. This paper gives an overview of the various cohorts, collected data, and most important research findings to date. In brief, the Framingham Heart Study, funded by the National Institutes of Health and managed by Boston University, spans 3 generations of well phenotyped White persons and 2 cohorts comprised of racial and ethnic minority groups. These cohorts are densely phenotyped, with extensive longitudinal follow-up, and they continue to provide us with important information on human cardiovascular and noncardiovascular physiology over the lifespan, as well as to identify major risk factors for cardiovascular disease. This paper also summarizes some of the more recent progress in molecular epidemiology and discusses the future of the study.

Does time heal all wounds? Life course associations between child welfare involvement and mortality in prospective cohorts from Sweden and Britain

Child welfare involvement reflects childhood adversity and is associated with increased adult mortality, but it remains unclear how this association changes over the life course. Drawing on the Stockholm Birth Cohort Multigenerational Study (Sweden) and the National Childhood Development Study (Great Britain) this study examines whether inequalities within these cohorts diverge or converge. Involvement with child welfare services (ICWS) is divided into two levels ('child welfare contact' and 'out-of-home care'). For each cohort, we quantify absolute health inequalities as differences in cumulative probabilities of death (18-58 years) and temporary life expectancy; and relative inequalities as hazard ratios in ten-year intervals and ratios of lifetime lost. Persistently, ICWS was associated with premature mortality. The strength of the association varied by age, sex and level of ICWS. Consistently across both countries, the most robust relationship was between out-of-home care and mortality, with statistically significant age-specific hazard ratios ranging between 1.8 and 3.4 for males and 1.8-2.1 for females. Child welfare contact that did not result in out-of-home placement showed less consistent results. Among females the mortality gap developed later compared to males. Estimates attenuate after controlling for family socioeconomic and other background variables but patterns remain intact. Our results show that absolute inequalities widen with increasing age, while relative inequalities might peak in early adulthood and then stabilize in midlife. The relative disadvantage among looked-after children in early adulthood is heightened by overall low rates of mortality at this age. Absolute inequality increases with age, highlighting the weight of the accumulation of disadvantage in mortality over time. The bulk of excess deaths that could be attributed to ICWS occurs from midlife onwards. Mechanisms that uphold the disadvantage after childhood experiences require further exploration. This study highlights that the association between out-of-home care and premature mortality seems to transcend welfare systems.

Role of cardiometabolic risk in the association between accumulation of affective symptoms across adulthood and mid-life cognitive function: national cohort study

BACKGROUND

Affective symptoms are associated with cognition in mid-life and later life. However, the role of cardiometabolic risk in this association has not been previously examined.

AIMS

To investigate how cardiometabolic risk contributes to associations between affective symptoms and mid-life cognition.

METHOD

Data were used from the National Child Development Study (NCDS), a sample of people born in Britain during one week in 1958. Measures of immediate and delayed memory, verbal fluency and information processing speed and accuracy were available at age 50. Affective symptoms were assessed at ages 23, 33 and 42 years and a measure of accumulation was derived. A cardiometabolic risk score was calculated from nine cardiometabolic biomarkers at age 44. Path models were run to test these associations, adjusting for sex, education, socioeconomic position and affective symptoms at age 50.

RESULTS

After accounting for missing data using multiple imputation, path models indicated significant indirect associations between affective symptoms and mid-life immediate memory (β = -0.002, s.e. = 0.001, P = 0.009), delayed memory (β = -0.002, s.e. = 0.001, P = 0.02) and verbal fluency (β = -0.002, s.e. = 0.001, P = 0.045) through cardiometabolic risk.

CONCLUSIONS

These findings suggest that cardiometabolic risk may play an important role in the association between affective symptoms and cognitive function (memory and verbal fluency). Results contribute to understanding of biological mechanisms underlying associations between affective symptoms and cognitive ageing, which can have implications for early detection of, and intervention for, those at risk of poorer cognitive outcomes.

Childhood socio-economic position and affective symptoms in adulthood: The role of neglect

BACKGROUND

Childhood neglect is more common within low-income families and can have long-term effects on mental health. Despite this, the extent to which it can mediate the well documented longitudinal inverse relationship between childhood socio-economic position (SEP) and adult affective symptoms is yet to be investigated.

METHOD

Data (9595 males and 8959 females) from participants of the National Child Development Study (NCDS) were used to investigate the extent to which prospectively measured neglect mediates the relationship between SEP (age 11) and affective symptoms (ages 23 and 50).

RESULTS

Neglect partially mediated the relationship between childhood SEP and affective symptoms at ages 23 (b = -0.02, [-0.02, -0.02]) and 50 (b = -0.02, [-0.02, -0.01]), after controlling for other family-related adversities. In addition, gender moderated the direct effect of SEP on affective symptoms at both ages 23 (b = -0.06, t = -4.87, [-0.08, -0.03]) and 50 (b = -0.05, t = -3.86, [-0.07, -0.02]), with the relationship being stronger for females; but did not moderate the indirect effect of neglect at either age 23 (b = 0.01, t = 1.09 [-0.01, 0.02]) or 50 (b = 0.00, t = -0.60 [-0.02, 0.01]).

CONCLUSIONS

Neglect in childhood should be viewed as having serious implications for the mental health of both men and women. Greater investments into social support interventions that reduce incidences of neglect are also warranted.

Changes over time in latent patterns of childhood-to-adulthood BMI development in Great Britain: evidence from three cohorts born in 1946, 1958, and 1970

BACKGROUND

Most studies on secular trends in body mass index (BMI) are cross-sectional and the few longitudinal studies have typically only investigated changes over time in mean BMI trajectories. We aimed to describe how the evolution of the obesity epidemic in Great Britain reflects shifts in the proportion of the population demonstrating different latent patterns of childhood-to-adulthood BMI development.

METHODS

We used pooled serial BMI data from 25,655 participants in three British cohorts: the 1946 National Survey of Health and Development (NSHD), 1958 National Child Development Study (NCDS), and 1970 British Cohort Study (BCS). Sex-specific growth mixture models captured latent patterns of BMI development between 11 and 42 years. The classes were characterised in terms of their birth cohort composition.

RESULTS

The best models had four classes, broadly similar for both sexes. The 'lowest' class (57% of males; 47% of females) represents the normal weight sub-population, the 'middle' class (16%; 15%) represents the sub-population who likely develop overweight in early/mid-adulthood, and the 'highest' class (6%; 9%) represents those who likely develop obesity in early/mid-adulthood. The remaining class (21%; 29%) reflects a sub-population with rapidly 'increasing' BMI between 11 and 42 years. Both sexes in the 1958 NCDS had greater odds of being in the 'highest' class compared to their peers in the 1946 NSHD but did not have greater odds of being in the 'increasing' class. Conversely, males and females in the 1970 BCS had 2.78 (2.15, 3.60) and 1.87 (1.53, 2.28), respectively, times higher odds of being in the 'increasing' class.

CONCLUSIONS

Our results suggest that the obesity epidemic in Great Britain reflects not only an upward shift in BMI trajectories but also a more recent increase in the number of individuals demonstrating more rapid weight gain, from normal weight to overweight, across the second, third, and fourth decades of life.

Multimorbidity and leisure-time physical activity over the life course: a population-based birth cohort study

BACKGROUND

We aimed to test which life course model best described the association between leisure-time physical activity (LTPA) and multimorbidity at age 55. We analyzed data from birth to age 55 using the database from the 1958 National Child Development Survey.

METHODS

Multimorbidity was considered as the presence of more than one chronic condition. LTPA was measured through questionnaires from 1965 (age 7) to 2013 (age 55), which were applied in eight different occasions. We compared the fit of a series of nested adjusted logistic regression models (representing either the critical, accumulation or sensitive period models) with a fully saturated model. Data were reported as odds ratio (OR) and 95% confidence interval (CI).

RESULTS

From an eligible sample of 15,613 cohort members, 9137 were interviewed in the latest sweep (58.5%). Men were more physically active than women at ages 11, 16, and 23 (p < 0.001). LTPA every day in the week was more frequent in women than men in ages 33, 42, and 50 (p < 0.001). The prevalence of multimorbidity at age 55 was 33.0% (n = 2778). The sensitive analysis revealed that LTPA during adolescence (OR: 0.83; 95% CI: 0.70, 0.98) and mid adult life (age 50 and 55; OR: 0.82; 95%CI: 0.69, 0.98) have a stronger effect on the risk for multimorbidity at age 55 considering all other life stages in the model. Also, adolescence showed a critical independent effect on the risk for multimorbidity (OR: 0.82; 95%CI: 0.70, 0.97). No difference was found between those models.

CONCLUSIONS

These data support the notion of a protective physical activity "legacy" at early ages of childhood against multimorbidity at older ages. We highlight the need for LTPA promotion through intervention tailored especially on schooling and older ages in order to reduce the burden of multimorbidity.

Disentangling trait, occasion-specific, and accumulated situational effects of psychological distress in adulthood: evidence from the 1958 and 1970 British birth cohorts

BACKGROUND

The trajectories of psychological distress differ between individuals, but these differences can be difficult to understand because the measures contain both consistent and situational features; however, in longitudinal studies these sources of information can be disentangled. In addition to occasion-specific features, interindividual differences can be decomposed into two sources of information: trait and carry-over effects between neighboring occasions that are not related to the trait (i.e. accumulated situational effects).

METHODS

To disentangle these three sources of variance throughout adulthood, the consistency (trait and accumulated situational effects) and occasion specificity of nine indicators of psychological distress from the Malaise Inventory were examined in two birth cohorts, the 1958 National Child Development Study (NCDS58), and the 1970 British Cohort Study (BCS70).

RESULTS

The scale was administered at ages 23, 33, 42, and 50 in NCDS58 (n = 7147), and at ages 26, 30, 34, and 42 in BCS70 (n = 6859). For each psychological symptom, more variance was consistent than occasion-specific. The majority of the consistency was due to trait variance as opposed to accumulated situational effects, indicating that an individual predisposed to be distressed at the beginning of the study remained more likely to be distressed over the whole period. Symptoms of rage were notably more consistent among males than females in both cohorts (78.1% and 81.3% variance explained by trait in NCDS58 and BCS70, respectively), and among females in the NCDS58 (69%).

CONCLUSIONS

Symptoms of psychological distress exhibited high stability throughout adulthood, especially among men, due mostly to interindividual trait differences.

Determinants of the population health distribution: an illustration examining body mass index

Most epidemiological studies examine how risk factors relate to average difference in outcomes (linear regression) or odds of a binary outcome (logistic regression); they do not explicitly examine whether risk factors are associated differentially across the distribution of the health outcome investigated. This paper documents a phenomenon found repeatedly in the minority of epidemiological studies which do this (via quantile regression): associations between a range of established risk factors and body mass index (BMI) are progressively stronger in the upper ends of the BMI distribution. In this paper, we document this finding and provide illustrative evidence of it in the 1958 British birth cohort study. Associations of low childhood socio-economic position, high maternal weight, low childhood general cognition and adult physical inactivity with higher BMI are larger at the upper end of the BMI distribution, on both absolute and relative scales. For example, effect estimates for socio-economic position and childhood cognition were around three times larger at the 90th compared with 10th quantile, while effect estimates for physical inactivity were increasingly larger from the 50th to 90th quantiles, yet null at lower quantiles. We provide potential explanations for these findings and discuss implications. Risk factors may have larger causal effects among those in worse health, and these effects may not be discovered when health is only examined in average terms. In such scenarios, population-based approaches to intervention may have larger benefits than anticipated when assuming equivalent benefit across the population. Further research is needed to understand why effect estimates differ across the BMI outcome distribution and to investigate whether differential effects exist for other physical and mental health outcomes.

Adult obesity and mid-life physical functioning in two British birth cohorts: investigating the mediating role of physical inactivity

Background

Associations between obesity and physical inactivity are bi-directional. Both are associated with physical functioning (PF, ability to perform physical tasks of daily living) but whether obesity influences PF via inactivity is unknown. We investigated whether mid-adult obesity trajectories were associated with subsequent PF and mediated by inactivity.

Methods

Body mass index (BMI; kg/m²) and inactivity were recorded at: 36, 43, 53 and 60–64 years in the 1946 Medical Research Council (MRC) National Survey of Health and Development (1946-NSHD; n = 2427), and at 33, 42 and 50 years in the 1958 National Child Development Study (1958-NCDS; n = 8674). Poor PF was defined as the lowest (gender and cohort-specific) 10% on the Short-form 36 Physical Component Summary subscale at 60–64 years (1946-NSHD) and 50 years (1958-NCDS). Estimated randomized-interventional-analogue natural direct (rNDE) and indirect (rNIE) effects of obesity trajectories on PF via inactivity are expressed as risk ratios [overall total effect (rTE) is rNDE multiplied by rNIE].

Results

In both cohorts, most individuals (∼68%) were never obese in adulthood, 16–30% became obese and ≤11% were always obese. In 1946-NSHD, rTE of incident obesity at 43 years (vs never) on poor PF was 2.32 (1.13, 3.51); at 53 years it was 1.53 (0.91, 2.15). rNIEs via inactivity were 1.02 (0.97, 1.07) and 1.02 (0.99, 1.04), respectively. Estimated rTE of persistent obesity from 36 years was 2.91 (1.14, 4.69), with rNIE of 1.03 (0.96, 1.10). In 1958-NCDS, patterns of association were similar, albeit weaker.

Conclusions

Longer duration of obesity was associated with increased risk of poor PF. Inactivity played a small mediating role. Findings reinforce the importance of preventing and delaying obesity onset to protect against poor PF.

Using a cross-cohort comparison design to test the role of maternal smoking in pregnancy in child mental health and learning: evidence from two UK cohorts born four decades apart

Background

Maternal smoking in pregnancy is associated with low birth weight (LBW), child conduct problems, hyperactivity and lower cognitive attainment, but associations may reflect measured and unmeasured confounding. Cross-cohort designs can aid causal inference through comparison of associations across populations with different confounding structures. We compared associations between maternal smoking in pregnancy and child conduct and hyperactivity problems, cognition and LBW across two cohorts born four decades apart.

Methods

Two national UK cohorts born in 1958 (n = 12 415) and 2000/01 (n = 11 800) were compared. Maternal smoking in pregnancy and child birth weight was assessed at or shortly after birth. Parents rated children’s conduct problems and hyperactivity, and children completed standardized tests of reading and mathematics.

Results

Maternal smoking in pregnancy was less common and more strongly associated with social disadvantage in 2000/01 compared with 1958 (interactions P < 0.001). Maternal smoking in pregnancy was robustly and equivalently associated with infant LBW in both cohorts [interactions: boys odds ratio (OR) = 1.01 (0.89, 1.16), P = 0.838; girls OR = 1.01 (0.91, 1.17), P = 0.633]. Maternal smoking was more strongly associated with conduct problems, hyperactivity and reading in the 2000/01 cohort (interactions P < 0.001).

Conclusions

Marked cross-cohort change in associations between maternal smoking and child conduct problems, hyperactivity and reading highlights the likely role of confounding factors. In contrast, association with LBW was unaffected by change in prevalence of maternal smoking and patterns of confounding. The study highlights the utility of cross-cohort designs in helping triangulate conclusions about the role of putative causal risk factors in observational epidemiology.

Birthweight, lifetime obesity and physical functioning in mid-adulthood: a nationwide birth cohort study

BACKGROUND

Evidence is scant on long-term implications of childhood obesity and body mass index (BMI) gains over the life-course for poor physical functioning (PF). The objective was to establish whether (i) birthweight and BMI across the life-course, (ii) BMI gains at specific life-stages and (iii) age of obesity onset were associated with PF at 50 y.

METHODS

In the 1958 British birth cohort (n = 8674), BMI (kg/m2) was calculated using height and weight [measured (7, 11, 16, 33 and 45 y); self-reported (23 and 50 y)]. PF was assessed at 50 y using the validated PF subscale of the Short-form 36 survey; the bottom (gender-specific) 10% was classified as poor PF. Missing data were imputed via multiple imputation. Associations were examined using logistic regression, adjusting for health and social factors.

RESULTS

Birthweight was not associated with PF. At each adult age, odds of poor PF were highest for obese (vs normal), e.g. for 23 y obesity the odds ratio (OR)adjusted for poor PF was 2.28 (1.34, 3.91) and 2.67 (1.72, 4.14) in males and females respectively. BMI gains were associated with poor PF, e.g. for females, ORadjusted per standard deviation (SD) in BMI gain 16-23 y was 1.28 (1.13, 1.46); for BMI gains 45-50 y it was 1.36 (1.11, 1.65). Longer duration of obesity was associated with poor PF, e.g. in males, ORadjusted was 2.32 (1.26, 4.29) for childhood obesity onset and 1.50 (1.16, 1.96) for mid-adulthood onset (vs never obese, P-trend < 0.001).

CONCLUSION

Obesity, BMI gains, and earlier obesity onset were associated with poor PF in mid-adulthood, reinforcing the importance of preventing and delaying obesity onset.

Mental health crisis in midlife – a proposed research agenda

There is a growing amount of evidence indicating increased levels of psychological distress, suicide rates and decreased well-being in midlife (age 45-55). We refer to this phenomenon as the ‘midlife mental health crisis’. As there is little empirical evidence or theoretical grounds to explain the midlife mental health crisis, we propose a research agenda.

In order to facilitate further research, we consulted members of public, mental health professionals and researchers on potential reasons for the midlife mental health crisis. Subsequently, we translated those into research questions testable with the British birth cohorts. We propose a series of studies using three statistical modelling approaches: descriptive (what is the midlife mental health crisis?), predictive (who is at increased risk of experiencing the midlife mental health crisis?) and explanatory (what are the processes leading to the midlife mental health crisis?).

Evaluating access to health and care services during lockdown by the COVID-19 survey in five UK national longitudinal studies

OBJECTIVE

Access to health services and adequate care is influenced by sex, ethnicity, socioeconomic position (SEP) and the burden of comorbidities. Our study aimed to assess whether the COVID-19 pandemic further deepened these already existing health inequalities.

DESIGN

Cross-sectional study.

SETTING

Data were collected from five longitudinal age-homogenous British cohorts (born in 2000-2002, 1989-1990, 1970, 1958 and 1946).

PARTICIPANTS

A web survey was sent to the cohorts. Anybody who responded to the survey was included, resulting in 14 891 eligible participants.

MAIN OUTCOMES MEASURED

The survey provided data on cancelled surgical or medical appointments, and the number of care hours received in a week during the first UK COVID-19 national lockdown.

INTERVENTIONS

Using binary or ordered logistic regression, we evaluated whether these outcomes differed by sex, ethnicity, SEP and having a chronic illness. Adjustment was made for study design, non-response weights, psychological distress, presence of children or adolescents in the household, COVID-19 infection, key worker status, and whether participants had received a shielding letter. Meta-analyses were performed across the cohorts, and meta-regression was used to evaluate the effect of age as a moderator.

RESULTS

Women (OR 1.40, 95% CI 1.27 to 1.55) and those with a chronic illness (OR 1.84, 95% CI 1.65 to 2.05) experienced significantly more cancellations during lockdown (all p<0.0001). Ethnic minorities and those with a chronic illness required a higher number of care hours during the lockdown (both OR≈2.00, all p<0.002). SEP was not associated with cancellation or care hours. Age was not independently associated with either outcome in the meta-regression.

CONCLUSION

The UK government's lockdown approach during the COVID-19 pandemic appears to have deepened existing health inequalities, impacting predominantly women, ethnic minorities and those with chronic illnesses. Public health authorities need to implement urgent policies to ensure equitable access to health and care for all in preparation for a fourthwave.

Evaluating access to health and care services during lockdown by the COVID-19 survey in five UK national longitudinal studies

OBJECTIVE

Access to health services and adequate care is influenced by sex, ethnicity, socioeconomic position (SEP) and the burden of comorbidities. Our study aimed to assess whether the COVID-19 pandemic further deepened these already existing health inequalities.

DESIGN

Cross-sectional study.

SETTING

Data were collected from five longitudinal age-homogenous British cohorts (born in 2000-2002, 1989-1990, 1970, 1958 and 1946).

PARTICIPANTS

A web survey was sent to the cohorts. Anybody who responded to the survey was included, resulting in 14 891 eligible participants.

MAIN OUTCOMES MEASURED

The survey provided data on cancelled surgical or medical appointments, and the number of care hours received in a week during the first UK COVID-19 national lockdown.

INTERVENTIONS

Using binary or ordered logistic regression, we evaluated whether these outcomes differed by sex, ethnicity, SEP and having a chronic illness. Adjustment was made for study design, non-response weights, psychological distress, presence of children or adolescents in the household, COVID-19 infection, key worker status, and whether participants had received a shielding letter. Meta-analyses were performed across the cohorts, and meta-regression was used to evaluate the effect of age as a moderator.

RESULTS

Women (OR 1.40, 95% CI 1.27 to 1.55) and those with a chronic illness (OR 1.84, 95% CI 1.65 to 2.05) experienced significantly more cancellations during lockdown (all p<0.0001). Ethnic minorities and those with a chronic illness required a higher number of care hours during the lockdown (both OR≈2.00, all p<0.002). SEP was not associated with cancellation or care hours. Age was not independently associated with either outcome in the meta-regression.

CONCLUSION

The UK government's lockdown approach during the COVID-19 pandemic appears to have deepened existing health inequalities, impacting predominantly women, ethnic minorities and those with chronic illnesses. Public health authorities need to implement urgent policies to ensure equitable access to health and care for all in preparation for a fourthwave.

Mental health crisis in midlife – a proposed research agenda

There is a growing amount of evidence indicating increased levels of psychological distress, suicide rates and decreased well-being in midlife (age 45-55). We refer to this phenomenon as the ‘midlife mental health crisis’. As there is little empirical evidence or theoretical grounds to explain the midlife mental health crisis, we propose a research agenda.

In order to facilitate further research, we consulted members of public, mental health professionals and researchers on potential reasons for the midlife mental health crisis. Subsequently, we translated those into research questions testable with the British birth cohorts. We propose a series of studies using three statistical modelling approaches: descriptive (what is the midlife mental health crisis?), predictive (who is at increased risk of experiencing the midlife mental health crisis?) and explanatory (what are the processes leading to the midlife mental health crisis?).

Association of public care in childhood with social, criminal, cognitive, and health outcomes in middle-age: five decades of follow-up of members of the 1958 birth cohort study

BACKGROUND

Children who have been exposed to public (out-of-home) care experience a range of negative outcomes by late adolescence and early adulthood. The longer-term impact of childhood care is, however, uncertain.

AIM

To examine if there is a prospective association between childhood public care and adverse life outcomes in middle-age.

METHODS

We used data from the UK 1958 birth cohort study of 18 558 individuals. Parents reported offspring care status at age 7, 11 and 16. An array of social, criminal, cognitive, and health outcomes was self-reported by cohort members at age 42 (71% response proportion in eligible sample) and a cognitive test battery was administered at age 50 (62% response).

RESULTS

A total of 420 (3.8%) of 11 160 people in the analytical sample experienced childhood public care by age 16. Net of confounding factors, experience of public care (vs none) was linked to 11 of the 28 non-mutually exclusive endpoints captured in middle-age, with the most consistent effects apparent for psychosocial characteristics: 4/7 sociodemographic (eg, odds ratio; 95% confidence interval for homelessness: 2.1; 1.4 to 3.1); 2/2 antisocial (eg, use of illicit drug: 2.0; 1.2 to 3.5); 2/3 psychological (eg, mental distress: 1.6; 1.2 to 2.1); 1/3 health behaviours (eg, current cigarette smoker: 1.7; 1.3 to 2.2); 2/8 somatic health (physical disability: 2.7; 1.9 to 3.8); and 0/5 cognitive function (eg, beta coefficient; 95% confidence interval for immediate word recall: -0.1; -0.3 to 0.1) endpoints.

CONCLUSIONS

The present study suggests that selected associations apparent between childhood care and outcomes in adolescence and early adulthood are also evident in middle-age.

Machine Learning Assessment of Early Life Factors Predicting Suicide Attempt in Adolescence or Young Adulthood

IMPORTANCE

Although longitudinal studies have reported associations between early life factors (ie, in-utero/perinatal/infancy) and long-term suicidal behavior, they have concentrated on 1 or few selected factors, and established associations, but did not investigate if early-life factors predict suicidal behavior.

OBJECTIVE

To identify and evaluate the ability of early-life factors to predict suicide attempt in adolescents and young adults from the general population.

DESIGN, SETTING, AND PARTICIPANTS

This prognostic study used data from the Québec Longitudinal Study of Child Development, a population-based longitudinal study from Québec province, Canada. Participants were followed-up from birth to age 20 years. Random forest classification algorithms were developed to predict suicide attempt. To avoid overfitting, prediction performance indices were assessed across 50 randomly split subsamples, and then the mean was calculated. Data were analyzed from November 2019 to June 2020.

EXPOSURES

Factors considered in the analysis included 150 variables, spanning virtually all early life domains, including pregnancy and birth information; child, parents, and neighborhood characteristics; parenting and family functioning; parents' mental health; and child temperament, as assessed by mothers, fathers, and hospital birth records.

MAIN OUTCOMES AND MEASURES

The main outcome was self-reported suicide attempt by age 20 years.

RESULTS

Among 1623 included youths aged 20 years, 845 (52.1%) were female and 778 (47.9%) were male. Models show moderate prediction performance. The areas under the curve for the prediction of suicide attempt were 0.72 (95% CI, 0.71-0.73) for females and 0.62 (95% CI, 0.60-0.62) for males. The models showed low sensitivity (females, 0.50; males, 0.32), moderate positive predictive values (females, 0.60; males, 0.62), and good specificity (females, 0.76; males, 0.82) and negative predicted values (females, 0.75; males, 0.71). The most important factors contributing to the prediction included socioeconomic and demographic characteristics of the family (eg, mother and father education and age, socioeconomic status, neighborhood characteristics), parents' psychological state (specifically parents' antisocial behaviors) and parenting practices. Birth-related variables also contributed to the prediction of suicidal behavior (eg, prematurity). Sex differences were also identified, with family-related socioeconomic and demographic characteristics being the top factors for females and parents' antisocial behavior being the top factor for males.

CONCLUSIONS AND RELEVANCE

These findings suggest that early life factors contributed modestly to the prediction of suicidal behavior in adolescence and young adulthood. Although these factors may inform the understanding of the etiological processes of suicide, their utility in the long-term prediction of suicide attempt was limited.

Missing at random assumption made more plausible: evidence from the 1958 British birth cohort

OBJECTIVE

Non-response is unavoidable in longitudinal surveys. The consequences are lower statistical power and the potential for bias. We implemented a systematic data-driven approach to identify predictors of non-response in the National Child Development Study (NCDS; 1958 British birth cohort). Such variables can help make the missing at random assumption more plausible, which has implications for the handling of missing data STUDY DESIGN AND SETTING: We identified predictors of non-response using data from the 11 sweeps (birth to age 55) of the NCDS (n = 17,415), employing parametric regressions and the LASSO for variable selection.

RESULTS

Disadvantaged socio-economic background in childhood, worse mental health and lower cognitive ability in early life, and lack of civic and social participation in adulthood were consistently associated with non-response. Using this information, along with other data from NCDS, we were able to replicate the "population distribution" of educational attainment and marital status (derived from external data), and the original distributions of key early life characteristics.

CONCLUSION

The identified predictors of non-response have the potential to improve the plausibility of the missing at random assumption. They can be straightforwardly used as "auxiliary variables" in analyses with principled methods to reduce bias due to missing data.

Childhood cognitive skill trajectories and suicide by mid-adulthood: an investigation of the 1958 British Birth Cohort

BACKGROUND

Poor cognitive abilities and low intellectual quotient (IQ) are associated with an increased risk of suicide attempts and suicide mortality. However, knowledge of how this association develops across the life-course is limited. Our study aims to establish whether individuals who died by suicide by mid-adulthood are distinguishable by their child-to-adolescence cognitive trajectories.

METHODS

Participants were from the 1958 British Birth Cohort and were assessed for academic performance at ages 7, 11, and 16 and intelligence at 11 years. Suicides occurring by September 2012 were identified from linked national death certificates. We compared mean mathematics and reading abilities and rate of change across 7-16 years for individuals who died by suicide v. those still alive, with and without adjustment for potential early-life confounding factors. Analyses were based on 14 505 participants.

RESULTS

Fifty-five participants (48 males) had died by suicide by age 54 years. While males who died by suicide did not differ from participants still alive in reading scores at age 7 [effect size (g) = -0.04, p = 0.759], their reading scores had a less steep improvement up to age 16 compared to other participants. Adjustments for early-life confounding factors explained these differences. A similar pattern was observed for mathematics scores. There was no difference between individuals who died by suicide v. participants still alive on intelligence at 11 years.

CONCLUSIONS

While no differences in tests of academic performance and IQ were observed, individuals who died by suicide had a less steep improvement in reading abilities over time compared to same-age peers.

Childhood correlates of adult positive mental well-being in three British longitudinal studies

BACKGROUND

Previous evidence has shown how experiences within childhood, such as parenting and socioeconomic conditions, are associated later on in life with adult mental well-being. However, these studies tend to focus on childhood experiences in isolation, and fewer studies have investigated how multiple aspects of the childhood environment, including both socioeconomic and psychosocial aspects, are associated with adult positive mental well-being. Using data from three British birth cohort studies, we investigated how prospective measures of the childhood environment up to the age of 16 years were associated with midlife adult mental well-being and whether similar associations were replicated across different generations.

METHODS

Childhood environment comprised socioeconomic circumstances, psychosocial factors (child-rearing and parenting, family instability) and parental health. The Warwick-Edinburgh Mental Wellbeing Scale, a validated instrument measuring both hedonic and eudaemonic aspects of well-being, was administered in mid-life. We modelled associations between childhood environment domains and well-being.

RESULTS

Despite changes in social context in all three studies, poorer quality parent-child relationships and poor parental mental health were strongly and independently associated with poorer adult mental well-being. Socioeconomic circumstances were also associated with adult mental well-being, but the association was weaker than for the measures of parenting or parental mental health.

CONCLUSION

These findings confirm that parenting and parental mental health, as well as socioeconomic circumstances, are important for adult mental well-being. Interventions in early childhood aimed at reducing socioeconomic adversity and offering support to parents might be warranted, to enhance adult mental well-being later on in the life course.

Socioeconomic inequalities across life and premature mortality from 1971 to 2016: findings from three British birth cohorts born in 1946, 1958 and 1970

INTRODUCTION

Disadvantaged socioeconomic position (SEP) in early and adult life has been repeatedly associated with premature mortality. However, it is unclear whether these inequalities differ across time, nor if they are consistent across different SEP indicators.

METHODS

British birth cohorts born in 1946, 1958 and 1970 were used, and multiple SEP indicators in early and adult life were examined. Deaths were identified via national statistics or notifications. Cox proportional hazard models were used to estimate associations between ridit scored SEP indicators and all-cause mortality risk-from 26 to 43 years (n=40 784), 26 to 58 years (n=35 431) and 26 to 70 years (n=5353).

RESULTS

More disadvantaged SEP was associated with higher mortality risk-magnitudes of association were similar across cohort and each SEP indicator. For example, HRs (95% CI) from 26 to 43 years comparing lowest to highest paternal social class were 2.74 (1.02 to 7.32) in 1946c, 1.66 (1.03 to 2.69) in 1958c, and 1.94 (1.20 to 3.15) in 1970c. Paternal social class, adult social class and housing tenure were each independently associated with mortality risk.

CONCLUSIONS

Socioeconomic circumstances in early and adult life show persisting associations with premature mortality from 1971 to 2016, reaffirming the need to address socioeconomic factors across life to reduce inequalities in survival to older age.

Premature and Early Menopause in Relation to Cardiovascular Disease

Postmenopausal women have an increased risk for cardiovascular diseases. It has been postulated that the loss of ovarian function and subsequent deficiency of endogenous estrogens after menopause contributes to this elevated risk of cardiovascular disease in postmenopausal women. Compared with woman entering menopause at the mean age of 51 years, in women with early menopause or premature ovarian insufficiency the risk for cardiovascular disease is even greater. These women lack the cardioprotective effect of endogenous estrogens for many more years than do women entering natural menopause. The majority of data assessing the risk of cardiovascular disease in relation to age at menopause and specifically premature menopause are derived from large epidemiological cohort studies. In addition, observations in women undergoing bilateral oophorectomy at an early age provide convincing evidence regarding association between early menopause or POI and the development of cardiovascular events and mortality. Moreover, genetic variants associated with earlier age at menopause have also been found to increase the risk of cardiovascular events in women. It has been substantiated that hormone replacement therapy (HRT) decreases the risk for ischemic heart disease and eliminates the increased cardiovascular disease mortality. It is therefore crucial to start HRT as soon as possible, particularly in women with premature ovarian insufficiency.

Toward a Working Definition of eCohort Studies in Health Research: Narrative Literature Review

Background

The wide availability of internet-connected devices and new sensor technologies increasingly infuse longitudinal observational study designs and cohort studies. Simultaneously, the costly and time-consuming nature of traditional cohorts has given rise to alternative, technology-driven designs such as eCohorts, which remain inadequately described in the scientific literature.

Objective

The aim of this study was to outline and discuss what may constitute an eCohort, as well as to formulate a first working definition for health researchers based on a review of the relevant literature.

Methods

A two-staged review and synthesis process was performed comparing 10 traditional cohorts and 10 eCohorts across the six core steps in the life cycle of cohort designs.

Results

eCohorts are a novel type of technology-driven cohort study that are not physically linked to a clinical setting, follow more relaxed and not necessarily random sampling procedures, are primarily based on self-reported and digitally collected data, and systematically aim to leverage the internet and digitalization to achieve flexibility, interactivity, patient-centeredness, and scalability. This approach comes with some hurdles such as data quality, generalizability, and privacy concerns.

Conclusions

eCohorts have similarities to their traditional counterparts; however, they are sufficiently distinct to be treated as a separate type of cohort design. The novelty of eCohorts is associated with a range of strengths and weaknesses that require further exploration.

Association of Early-Life Mental Health With Biomarkers in Midlife and Premature Mortality: Evidence From the 1958 British Birth Cohort

Importance

Early-life mental health is known to be associated with socioeconomic adversity and psychological distress in adulthood, but less is known about potential associations with biomarkers and mortality.

Objective

To investigate the association between early-life mental health trajectories with biomarkers in midlife and premature mortality.

Design, Setting, and Participants

This study used data from the British National Child Development Study, a population-based birth cohort. The initial sample of 17 415 individuals consisted of all infants born in Great Britain in a single week in 1958. Analysis began Feburary 2017 and ended May 2020.

Main Outcomes and Measures

Biomarkers collected at age 44 to 45 years were fibrinogen, C-reactive protein, glycated hemoglobin, high- and low-density lipoprotein cholesterol, forced expiratory volume, blood pressure, and waist-to-hip ratio. Information on all-cause mortality was available up to age 58 years and cause-specific mortality up to age 50 years.

Results

Biomarkers were analyzed from 9377 participants (age, 44-45 years, 4712 female [50.3%]) and mortality data from 15 067 participants (age, 58 years; 7379 female [49.0%]). A 4-group longitudinal typology of early-life conduct problems and affective symptoms was identified: (1) stable low, (2) teacher-identified adolescent onset, (3) moderate, and (4) stable high. Compared with the stable-low group, the stable-high (B, 2.308; 95% CI, 0.213-4.404) and teacher-identified adolescent-onset (B, 2.114; 95% CI, 0.725-3.503) groups had less favorable levels of fibrinogen in middle age. Effect modification was observed by sex (P < .005) such that compared with the stable-low group, differences in high-density lipoprotein and abdominal obesity were only observed in female individuals. The stable-high and teacher-identified adolescent-onset groups had elevated risk for all-cause mortality (hazard ratio, 1.878; 95% CI, 1.501-2.350 and hazard ratio, 1.412; 95% CI, 1.174-1.698). Psychopathology-associated mortality was higher in both groups but unintentional injuries-associated mortality only in the stable-high group.

Conclusions and Relevance

Experiencing affective symptoms and conduct problems in childhood and adolescence is associated with less favorable levels of biomarkers 28 years later and elevated risk of premature mortality. These findings, if causal and generalizable to younger cohorts, may imply that effective interventions on early-life mental health have the potential to shift the distribution of risk and improve population health.

Associations Between Hearing and Cognitive Abilities From Childhood to Middle Age: The National Child Development Study 1958

Previous cross-sectional findings indicate that hearing and cognitive abilities are positively correlated in childhood, adulthood, and older age. We used an unusually valuable longitudinal dataset from a single-year birth cohort study, the National Child Development Study 1958, to test how hearing and cognitive abilities relate to one another across the life course from childhood to middle age. Cognitive ability was assessed with a single test of general cognitive ability at age 11 years and again with multiple tests at age 50. Hearing ability was assessed, using a pure tone audiogram, in childhood at ages 11 and 16 and again at age 44. Associations between childhood and middle-age hearing and cognitive abilities were investigated using structural equation modelling. We found that higher cognitive ability was associated with better hearing (indicated by a lower score on the hearing ability variables); this association was apparent in childhood (r  =  -0.120, p <0.001) and middle age (r  =  -0.208, p <0.001). There was a reciprocal relationship between hearing and cognitive abilities over time: better hearing in childhood was weakly associated with a higher cognitive ability in middle age (β  =  -0.076, p  =  0.001), and a higher cognitive ability in childhood was associated with better hearing in middle age (β  =  -0.163, p <0.001). This latter, stronger effect was mediated by occupational and health variables in adulthood. Our results point to the discovery of a potentially life-long relationship between hearing and cognitive abilities and demonstrate how these variables may influence one another over time.

Evidence for a causal association between milk intake and cardiometabolic disease outcomes using a two-sample Mendelian Randomization analysis in up to 1,904,220 individuals

BACKGROUND

High milk intake has been associated with cardio-metabolic risk. We conducted a Mendelian Randomization (MR) study to obtain evidence for the causal relationship between milk consumption and cardio-metabolic traits using the lactase persistence (LCT-13910 C > T, rs4988235) variant as an instrumental variable.

METHODS

We tested the association of LCT genotype with milk consumption (for validation) and with cardio-metabolic traits (for a possible causal association) in a meta-analysis of the data from three large-scale population-based studies (1958 British Birth Cohort, Health and Retirement study, and UK Biobank) with up to 417,236 participants and using summary statistics from consortia meta-analyses on intermediate traits (N = 123,665-697,307) and extended to cover disease endpoints (N = 86,995-149,821).

RESULTS

In the UK Biobank, carriers of 'T' allele of LCT variant were more likely to consume milk (P = 7.02 × 10^-14). In meta-analysis including UK Biobank, the 1958BC, the HRS, and consortia-based studies, under an additive model, 'T' allele was associated with higher body mass index (BMI) (P_{meta-analysis} = 4.68 × 10^-12) and lower total cholesterol (TC) (P = 2.40 × 10^-36), low-density lipoprotein cholesterol (LDL-C) (P = 2.08 × 10^-26) and high-density lipoprotein cholesterol (HDL-C) (P = 9.40 × 10^-13). In consortia meta-analyses, 'T' allele was associated with a lower risk of coronary artery disease (OR:0.86, 95% CI:0.75-0.99) but not with type 2 diabetes (OR:1.06, 95% CI:0.97-1.16). Furthermore, the two-sample MR analysis showed a causal association between genetically instrumented milk intake and higher BMI (P = 3.60 × 10^-5) and body fat (total body fat, leg fat, arm fat and trunk fat; P < 1.37 × 10^-6) and lower LDL-C (P = 3.60 × 10^-6), TC (P = 1.90 × 10^-6) and HDL-C (P = 3.00 × 10^-5).

CONCLUSIONS

Our large-scale MR study provides genetic evidence for the association of milk consumption with higher BMI but lower serum cholesterol levels. These data suggest no need to limit milk intakes with respect to cardiovascular disease risk, with the suggested benefits requiring confirmation in further studies.

Machine learning for genetic prediction of psychiatric disorders: a systematic review

Machine learning methods have been employed to make predictions in psychiatry from genotypes, with the potential to bring improved prediction of outcomes in psychiatric genetics; however, their current performance is unclear. We aim to systematically review machine learning methods for predicting psychiatric disorders from genetics alone and evaluate their discrimination, bias and implementation. Medline, PsycInfo, Web of Science and Scopus were searched for terms relating to genetics, psychiatric disorders and machine learning, including neural networks, random forests, support vector machines and boosting, on 10 September 2019. Following PRISMA guidelines, articles were screened for inclusion independently by two authors, extracted, and assessed for risk of bias. Overall, 63 full texts were assessed from a pool of 652 abstracts. Data were extracted for 77 models of schizophrenia, bipolar, autism or anorexia across 13 studies. Performance of machine learning methods was highly varied (0.48-0.95 AUC) and differed between schizophrenia (0.54-0.95 AUC), bipolar (0.48-0.65 AUC), autism (0.52-0.81 AUC) and anorexia (0.62-0.69 AUC). This is likely due to the high risk of bias identified in the study designs and analysis for reported results. Choices for predictor selection, hyperparameter search and validation methodology, and viewing of the test set during training were common causes of high risk of bias in analysis. Key steps in model development and validation were frequently not performed or unreported. Comparison of discrimination across studies was constrained by heterogeneity of predictors, outcome and measurement, in addition to sample overlap within and across studies. Given widespread high risk of bias and the small number of studies identified, it is important to ensure established analysis methods are adopted. We emphasise best practices in methodology and reporting for improving future studies.

The effect of exposure to long working hours on alcohol consumption, risky drinking and alcohol use disorder: A systematic review and meta-analysis from the WHO/ILO Joint Estimates of the Work-related Burden of Disease and Injury

BACKGROUND

The World Health Organization (WHO) and the International Labour Organization (ILO) are developing Joint Estimates of the work-related burden of disease and injury (WHO/ILO Joint Estimates), with contributions from a large network of experts. Evidence from mechanistic data suggests that exposure to long working hours may increase alcohol consumption and cause alcohol use disorder. In this paper, we present a systematic review and meta-analysis of parameters for estimating the number of deaths and disability-adjusted life years from alcohol consumption and alcohol use disorder that are attributable to exposure to long working hours, for the development of the WHO/ILO Joint Estimates.

OBJECTIVES

We aimed to systematically review and meta-analyse estimates of the effect of exposure to long working hours (three categories: 41-48, 49-54 and ≥55 h/week), compared with exposure to standard working hours (35-40 h/week), on alcohol consumption, risky drinking (three outcomes: prevalence, incidence and mortality) and alcohol use disorder (three outcomes: prevalence, incidence and mortality).

DATA SOURCES

We developed and published a protocol, applying the Navigation Guide as an organizing systematic review framework where feasible. We searched electronic bibliographic databases for potentially relevant records from published and unpublished studies, including the WHO International Clinical Trials Register, Ovid MEDLINE, PubMed, Embase, and CISDOC on 30 June 2018. Searches on PubMed were updated on 18 April 2020. We also searched electronic grey literature databases, Internet search engines and organizational websites; hand-searched reference list of previous systematic reviews and included study records; and consulted additional experts.

STUDY ELIGIBILITY AND CRITERIA

We included working-age (≥15 years) workers in the formal and informal economy in any WHO and/or ILO Member State but excluded children (<15 years) and unpaid domestic workers. We considered for inclusion randomized controlled trials, cohort studies, case-control studies and other non-randomized intervention studies with an estimate of the effect of exposure to long working hours (41-48, 49-54 and ≥55 h/week), compared with exposure to standard working hours (35-40 h/week), on alcohol consumption (in g/week), risky drinking, and alcohol use disorder (prevalence, incidence or mortality).

STUDY APPRAISAL AND SYNTHESIS METHODS

At least two review authors independently screened titles and abstracts against the eligibility criteria at a first stage and full texts of potentially eligible records at a second stage, followed by extraction of data from publications related to qualifying studies. Two or more review authors assessed the risk of bias, quality of evidence and strength of evidence, using Navigation Guide and GRADE tools and approaches adapted to this project.

RESULTS

Fourteen cohort studies met the inclusion criteria, comprising a total of 104,599 participants (52,107 females) in six countries of three WHO regions (Americas, South-East Asia, and Europe). The exposure and outcome were assessed with self-reported measures in most studies. Across included studies, risk of bias was generally probably high, with risk judged high or probably high for detection bias and missing data for alcohol consumption and risky drinking. Compared to working 35-40 h/week, exposure to working 41-48 h/week increased alcohol consumption by 10.4 g/week (95% confidence interval (CI) 5.59-15.20; seven studies; 25,904 participants, I2 71%, low quality evidence). Exposure to working 49-54 h/week increased alcohol consumption by 17.69 g/week (95% confidence interval (CI) 9.16-26.22; seven studies, 19,158 participants, I2 82%, low quality evidence). Exposure to working ≥55 h/week increased alcohol consumption by 16.29 g/week (95% confidence interval (CI) 7.93-24.65; seven studies; 19,692 participants; I2 82%, low quality evidence). We are uncertain about the effect of exposure to working 41-48 h/week, compared with working 35-40 h/week on developing risky drinking (relative risk 1.08; 95% CI 0.86-1.36; 12 studies; I2 52%, low certainty evidence). Working 49-54 h/week did not increase the risk of developing risky drinking (relative risk 1.12; 95% CI 0.90-1.39; 12 studies; 3832 participants; I2 24%, moderate certainty evidence), nor working ≥55 h/week (relative risk 1.11; 95% CI 0.95-1.30; 12 studies; 4525 participants; I2 0%, moderate certainty evidence). Subgroup analyses indicated that age may influence the association between long working hours and both alcohol consumption and risky drinking. We did not identify studies for which we had access to results on alcohol use disorder.

CONCLUSIONS

Overall, for alcohol consumption in g/week and for risky drinking, we judged this body of evidence to be of low certainty. Exposure to long working hours may have increased alcohol consumption, but we are uncertain about the effect on risky drinking. We found no eligible studies on the effect on alcohol use disorder. Producing estimates for the burden of alcohol use disorder attributable to exposure to long working hours appears to not be evidence-based at this time. PROTOCOL IDENTIFIER: https://doi.org/10.1016/j.envint.2018.07.025.

PROSPERO REGISTRATION NUMBER

CRD42018084077.

Empirical Evaluation of Pre-trained Transformers for Human-Level NLP: The Role of Sample Size and Dimensionality

In human-level NLP tasks, such as predicting mental health, personality, or demographics, the number of observations is often smaller than the standard 768+ hidden state sizes of each layer within modern transformer-based language models, limiting the ability to effectively leverage transformers. Here, we provide a systematic study on the role of dimension reduction methods (principal components analysis, factorization techniques, or multi-layer auto-encoders) as well as the dimensionality of embedding vectors and sample sizes as a function of predictive performance. We first find that fine-tuning large models with a limited amount of data pose a significant difficulty which can be overcome with a pre-trained dimension reduction regime. RoBERTa consistently achieves top performance in human-level tasks, with PCA giving benefit over other reduction methods in better handling users that write longer texts. Finally, we observe that a majority of the tasks achieve results comparable to the best performance with just 1 12 of the embedding dimensions.