Mapping domains of early-life determinants of future multimorbidity across three UK longitudinal cohort studies

Abstract

Many studies use a reductionist approach to isolate the influence of one factor in childhood on multimorbidity rather than consider the combined effect of wider determinants. We explored how multiple early-life determinants of multimorbidity can be characterised across three UK cohort studies. We used the National Child Development Study (NCDS), the 1970 British Cohort Study (BCS70), and the Aberdeen Children of the 1950s Study (ACONF) to identified early-life variables that fit into 12 domains of early-life determinants of multimorbidity. Variables were assigned into 12 domains; principal component analysis reduced the dimensionality of the data and structured the variables into subgroups. The data audit identified 7 domains in ACONF, 10 domains in NCDS and 12 domains in BCS70. Components included maternal fertility histories within the prenatal, antenatal and birth domain, long-term illnesses within the child health domain, educational ability within the child education and health literacy domain, ethnicity within the demography domain, parental health behaviours within the transgenerational domain, housing within the socioeconomic domain and parental-child interactions within the parental-family domain. Conceptualising the risk of future multimorbidity as lifecourse domains composed of multiple factors can help challenge the existing understanding of disease aetiology and develop new ideas for prevention of multimorbidity.

Exploring the Relationship Between Early Life Exposures and the Comorbidity of Obesity and Hypertension: Findings from the 1970 The British Cohort Study (BCS70)

“…Secondly, it is important to expand the methods presented here to consider the relationship to other multimorbidity outcomes that additionally explore the burdensome dimension of multimorbidity. Finally, in previous research [27] we audited the early life /variables available in two other cohort datasets. The data available in these datasets provide the opportunity to validate and compare our results.…”

“…Previously we developed a conceptualisation of 12 domains of early life risk factors of future multimorbidity risk informed through a scoping literature/policy review and patient and public engagement [26], and a data audit of these domains [27]. In this paper we focus our analysis on 5 out of the 12 domains chosen because they showed unadjusted associations with the outcomes:…”

“…Supplementary Materials Table 1 includes all the variables that were initially considered for each domain. These variables were selected and categorised based on a previous data audit and PCA analysis [27] that identified early-life variables from multiple sweeps of data that fitted into five domains of early-life domains of future multimorbidity risk. This previous work reduced the dimensionality of the data and structured each of the five domains into mutually exclusive groups of variables based on similar characteristics [26,27].…”

“…Many wider determinants acting in childhood are likely to increase the risk of disease in adulthood, and in previous research [26,27] we identified exposures across five pre-defined childhood domains covering a range of social, economic, developmental, educational and environmental factors. Yet, most previous research focuses on single exposure-outcome relationships, potentially to reduce statistical complexity, or to focus policy attention onto a specific aspect.…”

Cohort profile for the creation of the SAIL MELD-B e-cohort (SMC) and SAIL MELD-B children and Young adult e-cohort (SMYC)

“…achieved through the analysis of birth cohorts and routinely-collected electronic health record (EHR) data sources, using a combination of Artificial-Intelligence (AI)-enhanced epidemiological analysis and statistical methods. For a more detailed description of the MELD-B objectives and structure, see [35][36][37].…”