Size at birth and cognitive ability in late life: A systematic review

Introduction Recent evidence suggests that growth restriction in utero may lead to neurocognitive disorders in late life, either through impaired brain development or adverse metabolic programming. Methods Systematic review of literature investigating the relationship between size at birth and cognitive abilities in late life. The search, data extraction, and rating for the quality of reporting were conducted independently by two researchers. Results Of 533 selected studies, 11 were included in this systematic review and 10 of these were from high‐income setting. Of these 11 studies, eight indicated that lower birth weight is a risk factor for lower cognitive function in late life, at least in high‐income countries. The reported effect sizes were small and it was not possible to conduct meta‐analyses because of clinical heterogeneity Discussion A modest association of lower birth weight with lower cognitive abilities in late life is consistent with persisting effects of the prenatal environment on brain function. As with all observational studies, confounding is an alternative explanation. Further studies are required to elucidate the mechanisms.

hypothesis with two plausible pathways to cognitive ageing: (a) by a direct effect of reduced intrauterine nutrition (reflected in birth size) on fetal brain development leading to reduced cognitive reserve and decreased cognitive ability or (b) programming of metabolism in very early life by under-nutrition, leading to increased risk mediated through cardiometabolic disorders. 3 Quality of nutrition during intrauterine development, reflected crudely in size at birth, is an important determinant of lifelong function, health, and disease risk. 4 Birth weight and head circumference at birth are indicators of intrauterine growth and brain development, respectively. 5 Larger birth weight, the most widely researched birth size measure, is associated with better cognitive function and higher intelligence from infancy through the third decade of life in several populations and countries independent of social background. [6][7][8] This association of birth weight with cognition occurs across the whole spectrum of birth weight rather than being confined to an extreme group. However, the strength of this association is known to diminish as individuals reach middle age, and associations with growth in early life may not persist beyond midlife. 8 In a systematic review conducted in 2015, Grove and colleagues examined the relationship between birth weight and general cognitive ability in non-clinical adult populations. 8 This included 1 122 858 participants aged between 18 to 78.4 years from 19 studies. Of these, only eight could be included in a random-effects meta-analysis and three were in those aged 60 yrs and above. There was a modest association of birth weight with cognitive ability; with each kilogramme increase in birth weight, there was a 0.13 SD increase in general intelligence (95% CI, 0.07,-0. 19) in those aged less than 60 yrs, independent of gestational age and parental social class at birth. However, the effect size was much lower and not statistically significant in those aged 60 years and above (0.07 SD; 95% CI, −0.02 to 0.16). In addition to the small number of studies, the authors did not consider other birth size parameters (like head circumference, length at birth, and ponderal index), which are known to be associated with cognitive ability in this age group. [9][10][11] While birth weight was not a reliable predictor of cognitive ability or decline beyond midlife in this review, it would be premature to conclude that prenatal environment is not associated with cognitive ability in late life.
(both for individual domains and composite scores), and any other relevant key data. The quality of eligible studies was evaluated using the Strengthening the Reporting of OBservational Studies in Epidemiology (STROBE) checklist. 13 Two independent researchers (M.K. and S.J.) undertook data extraction and quality assessment. Disagreements were resolved by consensus.
If it was feasible to conduct a meta-analysis, it was planned to provide an estimate of combined effect size. If sufficient numbers of eligible studies were retrieved, it was planned to evaluate publication bias by a funnel plot analysis.

| Setting and design
The studies were published between 1996 and 2014 and included community-dwelling men and women who volunteered to participate.
Two studies had a cross-sectional design 11,14 while others were longitudinal follow-ups of established cohorts. 9,10,[15][16][17][18][19][20][21] Of the 11 studies, nine were cohort studies in which participants were matched to their birth records. The other two were community-based cohorts from the United States. Set up for examining cardiovascular disorders and birth weight was self-reported by the participants (Table 1).

| Demographics
The sample size ranged from 130 to 6875 and participants were aged

| Factors at birth
Birth weight was a universally available measurement of birth size across all the studies. In two studies 14,15 both from the USA, birth weight was obtained by recall and non-hospital records (such as family diaries and birth certificates), and did not provide any other information related to birth. All other studies were based on the birth weight obtained from obstetric records. As a measurement of birth size, only birth weight was available from obstetric records in Hyvarinen et al, while Muller et al had an additional measurement of length at birth.
In addition to birth weight, length at birth, head circumference, and gestational age were available from the maternity records in other studies. 9-11,16-18  Parental occupation as an indicator of socioeconomic position at birth was available from obstetric records in some studies, 9,11,16,19 while occupation of the head of the household was available from maternity records from de Rooij 2010 et al. Information about parental education at birth, an important determinant of growth and development of the offspring, was not available in any of the studies.

| Cognitive outcomes
All studies examined memory and attention, while most studies (n = 9) had a measure of verbal fluency as cognitive outcomes (Tables 1 and   2). Additional cognitive domains were examined in most of the studies.

| Confounding factors
The association of birth size with cognitive outcomes was adjusted for a range of confounding factors in most of the studies (Table 2). They include: gestational age, maternal age and parity, indicators of socioeconomic position at birth, attained educational level, social class of participants, and cardiometabolic risk factors. However, these studies do not provide information as to why these factors were thought to be confounding and/or were important as covariates.

| Estimates of effect sizes and analyses
The strength of association of birth size parameters with cognitive outcomes was examined and reported differently across studies (Table 2). In addition, many of the eligible studies were relatively small; from diverse population groups, both exposures and outcome measures for cognitive function were multiple and heterogeneous (Table 2). Therefore, it was not possible to conduct a metanalysis or evaluate for publication bias.   for individual studies is provided in Birth weight (self reported) was directly associated with cognitive abilities in this study. However, the authors were unable to provide data specific to those aged 50 yrs and above. Cohort. 25 Participants were aged between 40 to 66 yrs, and birth weight was self-reported. Authors were contacted and they were unable to provide data specific to those aged 50 yrs and above. The association of birth size with late life cognition was independent of parental socioeconomic position at birth in most studies [15][16][17][18][19] and was confounded by socioeconomic position at birth in one study. 11 Parental socioeconomic position at birth was not associated with cognitive function in late life in Shenkin et al, while this association was not examined in the remaining studies. 9,14,20,21 Across all the studies, adjusting for education attenuated the strength of association of birth size with late life cognition. When reported separately, higher level of attained education was directly associated with higher scores for certain cognitive abilities. Most studies in this review conducted baseline cognitive assessments when participants were well above the age of 50, by which cognitive decline may already be evident and observed associations (or a lack of) in these studies may be due to a horse racing effect. 30

| Quality of reporting and risk of bias
While examining cognitive function in the studies included in this review, cognitive decline may have been measured, and mostly the papers were uninformative about this. However, cognitive decline was specifically measured in four studies in this review. Of those that examined the relationship between birth size and cognitive decline, no association was reported in three studies, 9,10,16 while in one study, 17 men with larger size at birth had slower rate of cognitive decline in late life.
Findings from this review also suggest that the relationship of growth and development in early life is more likely with cognitive abilities that are associated with the fronto-temporal lobes of the brain such as verbal fluency, attention, trail making, calculation, executive functioning, and working memory. Of these, verbal fluency is regarded as an index of crystallised intelligence, while others are generally considered as components of fluid intelligence. In fact, the verbal fluency test is particularly sensitive to linguistic impairment and early mental decline in older persons; it is also a sensitive indicator of damage to the left lateral lobe. 31,32

| Strengths and limitations of included studies
The studies that reported a positive association of size at birth with late life cognitive ability generally included relatively well-educated, predominantly white, and middle-class men and women from higher income settings, which limits the generalisability of the findings beyond these settings. Moreover, the estimates of the effect sizes reported are at best modest. It is possible that the results are specific to the cohorts under study (cohort bias). These individuals have seen substantial changes in both prenatal and later health care.
None of the studies included in the review have information of the entire eligible population to assess the degree of potential bias. The studies used volunteers, who generally have higher cognitive ability and social class than non-volunteers. 33 As all analyses were performed within the study sample, unless the correlation between birth size and cognitive ability differs between the volunteers and non volunteers, it is unlikely that significant bias would have been introduced. Birth Participant exclusion is also known to introduce bias. Although most studies in this review excluded a minimal number of participants (Table 1), one study 15 excluded 36.6% (n = 3921) of participants examined at the initial visit and such an extensive exclusion may limit generalisability to the wider population. In two of the studies, 14,15 both from the United States, birth weight was obtained by recall Five studies [14][15][16]19,21 did not adjust the analyses for gestational age, which reduces the specificity of birth weight as a measure of fetal growth. This may have resulted in the lack of associations in some of these studies. Most studies did not provide justifications for the majority of adjustments (Table 2). Furthermore, one study 15 adjusted for a total of 21 different measures (not including gestational age), which makes it difficult to assess how far participants represent the general population. Some studies also did not provide any unadjusted information, making it difficult to assess the role of covariates in the reported effect (Table 2). Depression is related to both size at birth and cognitive function, 34 but the confounding effect of depression on the association of size at birth with cognitive ability was measured only in one study 20 in this review.
In this review, most studies report associations of multiple param-

| Strengths and limitations of the review process
This review strictly adhered to the study protocol, which was developed prior to the formal search. The forward citation search and reference list search were conducted systematically. Several authors of potentially eligible study were contacted for additional information.
All relevant studies appear to have been included in this review. There were no restrictions on publication language, and full-texts of all providing the full articles and assisting in this review.

CONFLICT OF INTEREST
None of the authors declare any conflict of interest.

AUTHOR CONTRIBUTIONS
The study was designed by M.

5-6
Key results 18 Summarise key results with reference to study objectives. 6-7

Limitations 19
Discuss limitations of the study, taking into account sources of potential bias or imprecision. Discuss both direction and magnitude of any potential bias.

6-7
Interpretation 20 Give a cautious overall interpretation of results considering objectives, limitations, multiplicity of analyses, results from similar studies, and other relevant evidence.

Other information
Funding 22 Give the source of funding and the role of the funders for the present study and, if applicable, for the original study on which the present article is based. 8

Item No Recommendation Page Number
Title and abstract 1 (a) Indicate the study's design with a commonly used term in the title or the abstract. 1

Study design 4
Present key elements of study design early in the paper. 2

Setting 5
Describe the setting, locations, and relevant dates, including periods of recruitment, exposure, follow-up, and data collection.
2 Participants 6 (a) Cohort study. Give the eligibility criteria and the sources and methods of selection of participants. Describe methods of follow-up. Case-control study. Give the eligibility criteria and the sources and methods of case ascertainment and control selection. Give the rationale for the choice of cases and controls. Cross-sectional study. Give the eligibility criteria and the sources and methods of selection of participants. Outcome data 15* Cohort study. Report numbers of outcome events or summary measures over time.

2-3
Case-control study. Report numbers in each exposure category or summary measures of exposure. Cross-sectional study. Report numbers of outcome events or summary measures.
Main results 16 (a) Give unadjusted estimates and, if applicable, confounderadjusted estimates and their precision (eg, 95% confidence interval). Make clear which confounders were adjusted for and why they were included.

2-3
(b) Report category boundaries when continuous variables were categorised. (c) If relevant, consider translating estimates of relative risk into absolute risk for a meaningful time period.
Other analyses 17 Report other analyses done-eg, analyses of subgroups and interactions and sensitivity analyses.

2-3
Discussion Key results 18 Summarise key results with reference to study objectives. 2

Limitations 19
Discuss limitations of the study, taking into account sources of potential bias or imprecision. Discuss both direction and magnitude of any potential bias.