Wild blueberries and cardiometabolic health: a current review of the evidence.

On any given day, millions of children worldwide experience parental incarceration (PI), and the number is growing as incarceration rates continue to rise globally, including the Western world (Clegg, Spitz, Usmani, & Wolcke, ). There is also a growing interest in the intergenerational effects of PI, with many studies reporting associations between PI and offspring internalizing problems, self‐harm, externalizing and risk‐taking behavior, and substance misuse (Boch & Ford, ; Luk, Hui, Tsang, Fung, & Chan, ; Rowell‐Cunsolo, Bellerose, & Borbely, ). However, individual studies vary considerably in the strength of these associations, reflecting substantial methodological heterogeneity across studies, such as variation in study populations, definitions, measures, and study designs (Giordano & Copp, ; Haskins, Amorim, & Mingo, ). One of the main unresolved issues concerns the nature of the observed associations. Most theoretical accounts assume a causal relationship between PI and offspring outcomes. For example, the risk of internalizing problems and self‐harm has been suggested to be explained by disrupted child–parent attachment after separation from the incarcerated parent, altered neurobiological stress responses following the traumatic loss, social stigma, and reduced family resources leading to increased parental stress, anxiety, and challenges to meet the child's emotional needs (Antle, Gibson, & Krohn, ; Beresford, Loucks, & Raikes, ; Johnson & Arditti, ; Luk et al., ). Explanations proposed for offspring externalizing behavior include social learning, reduced parental supervision accompanied with increased time spent with antisocial peers, and the adoption of such behaviors as a coping mechanism (Antle et al., ; Johnson & Arditti, ; Rowell‐Cunsolo et al., ). However, the associations may also be partly or fully explained by other disadvantage and hardship often present in these families. Children experiencing PI are a highly selected group exposed to a range of other adversities, such as parental crime, mental disorders, substance misuse, family economic hardship, domestic violence, and child abuse (Boch, Warren, & Ford, ; Giordano & Copp, ; Turney, ). Also, shared genetic liabilities between parents and children may contribute to elevated risks (Murray, Farrington, & Sekol, ). Accounting for the impact of clustered disadvantage constitutes a central methodological challenge in this area of research (Giordano & Copp, ; Haskins et al., ; Kirk & Wakefield, ). The ability to account for selection effects varies widely across studies, contributing to heterogeneity in their findings. Most existing studies have estimated associations between PI and offspring outcomes by comparing unrelated individuals who differ in their exposure to PI, adjusting for observed covariates to mitigate selection bias. In such studies, PI has typically been associated with offspring outcomes (e.g., Alcalá, Horino, & Delva, ; Bravo et al., ; Gifford et al., ; Le, Deardorff, Lahiff, & Harley, ). Some studies have made greater efforts to address causality, for example, by extensively adjusting for between‐group differences in numerous background characteristics and childhood adversities, often observing a reduction in the relative importance of PI in predicting offspring outcomes (e.g., Boch et al., ; Turney, ). Other studies have used clinic or court samples or comparison groups consisting of other disadvantaged children to better isolate the potential independent effect of PI from that of background factors, often finding attenuated associations (Jackson, Testa, Semenza, & Vaughn, ; Murray et al., ; Murray, Bijleveljd, Farrington, & Loeber, ). A few studies have relied on propensity score matching methods aiming to ensure balanced unexposed and exposed groups by matching individuals on measured background factors. These studies have provided somewhat heterogenous results, with associations attenuating to non‐significance in some studies (e.g., Bradshaw, Hannigan, Creaven, & Muldoon, ; Turney & Wildeman, ), but remaining significant in others (e.g., Mears & Siennick, ; Turney, ). While studies that more carefully consider confounding suggest somewhat less consistent associations between PI and offspring health and behavioral outcomes, they have been largely unable to consider selection bias caused by unobserved confounding, such as genetic and unmeasured environmental factors. This limits their ability to fully answer the key question of whether PI is a risk marker or an independent risk factor.