Parental imprisonment as a risk factor for cardiovascular and metabolic disease in adolescent and adult offspring: A prospective Australian birth cohort study

Objectives: Parental imprisonment is linked with child health in later life. The present study provides the first prospective cohort analysis and non-U.S. based study examining parental imprisonment and cardiometabolic risk factors in adolescence and adulthood. Methods: The study followed 7,223 children born from live, singleton births from 1981 to 1984 in Brisbane, Australia. Data on parental imprisonment was collected at mother interview when the children were ages 5 and 14. Our sample analyzes offspring with biometric data collected by health professionals, including 3,794 at age 14, 2,136 at age 21, and 1,712 at age 30. Analyses used multivariate linear and logistic regression, and time-varying growth curve models. Results: Among female respondents, parental imprisonment at ages ≤5 was associated with higher body-mass index (BMI) at ages 14, 21, and 30; higher systolic blood pressure (SBP) and diastolic blood pressure (DBP) at age 30; and increased sedentary hours, larger waist circumference, and odds of a high-risk waist circumference at age 30. Parental imprisonment when the child was aged ≤14 was associated with increased BMI and SBP at age 30 for females. In growth-curve models, parental imprisonment when the child was aged ≤5 and ≤ 14 among females was linked with increased BMI; parental imprisonment when the child was aged ≤5 was associated with increased SBP and DBP. No significant associations were observed for males. Conclusions: Using prospective cohort data, our results support research showing that parental imprisonment, particularly in early childhood, is associated with increased BMI, blood pressure, sedentary hours, and waist circumference in females in early adulthood. These findings implicate parental imprisonment as a risk factor for cardiometabolic health issues in later life among females. We use data from the Mater Hospital-University of Queensland Study on Pregnancy (MUSP). The MUSP is a cohort study of 7,223 mothers whose pregnancies resulted in live, singleton births from 1981 to 1984 in the obstetrics unit of the Mater-Misericordiae Hospital in Brisbane, Australia. Mother and child clinical and survey data have been collected at several follow-up waves until the children reached age 30. This study uses data collected from mothers prenatally and when their children were aged 5 and 14; and child data are from waves of data collected at ages 14, 21, and 30. Among the 7,223 children who were initially enrolled in the study, we examine subsets of respondents who had clinical biometric data completed at ages 14, 21, and 30 during physical assessments by a trained health professional. These numbers include 3,794 respondents at age 14, 2,136 respondents at age 21, and 1,712 respondents at age 30. Further details of the MUSP data are available in the MUSP cohort profiles and research publications (Najman et al., 2005, 2015). The attrition in the sample is a potential issue for the representativeness of the data. As noted in two MUSP cohort profiles, early and later attrition through age 30 have not been found to substantially bias results for parent and child outcomes in the MUSP and the representativeness of the original Brisbane sample (Najman et al., 2005, 2015). Attrition analyses of the MUSP data for mothers found that having problems with the law was not associated with increased attrition, while ethnic minorities and those with lower SES were more likely be lost in the sample (Saiepour et al., 2019); to address this potential attrition bias, we control for family SES and ethnicity in the analysis. Attrition for biometric measures, such as blood pressure, height, and weight, has been limited due to collection of these measures at the Mater Misericordiae clinic, limiting access for those who may have moved outside of the Brisbane area (Das et al., 2020). However, this limitation for the biometric data has not been found to bias results using the age 21 and age 30 cohort data in research (Das et al., 2020; Najman et al., 2020). Ethics approval was received from relevant committees at The University of Queensland and the Mater Misericordiae Hospital, South Brisbane, Australia for data collection. For the present study, we use deidentified secondary data exempt from Human Ethics approval. To maintain confidentiality, data from the MUSP are not made publicly available. Data may be obtained from the University of Queensland through the study website at: https://social-science.uq.edu.au/mater-university-queensland-study-pregnancy?p=9#9. Body mass index (BMI, kg/m2). Based on measured height (meters) and weight (kilograms) during physical assessments at ages 14, 21, and 30. Normal body mass is in the range 18.5 ≤ BMI<25, overweight BMI is in range 25 ≤ BMI<30, and obesity is BMI≥30 (Conen et al., 2007). Systolic Blood Pressure (SBP). SBP (mmHG) at ages 21 and 30 was measured during physical assessments. Two readings were taken 5 min apart when the respondents were sitting and at rest. The respondent's SBP was the average of these two readings (Das et al., 2020). For SBP, categories are optimal for readings less than120 mmHG, normal is 120–129 mmHG, high-normal is 130–139 mmHG, and hypertension is ≥ 140 mmHG (Conen et al., 2007). Diastolic Blood Pressure (DBP). DBP (mmHG) at ages 21 and 30 was measured during physical assessments. Two readings were taken 5 min apart when the respondents were sitting and at rest. The respondent's DBP was the average of these two readings (Das et al., 2020). For DBP, optimal blood pressure is > 75 mmHG, normal is 74–84, high-normal is 85–89 mmHG, and hypertension is ≥ 90 mmHG (Conen et al., 2007). Sedentary hours. Number of self-reported hours per day over the prior week spent watching TV or using a computer for non-work purposes at age 30. Increased sedentary time, such as TV viewing time, is linked to increased cardiovascular risk (Wijndaele et al., 2010). Waist size. Self-reported waist size (centimetres) at age 30. To measure waist size, respondents were provided with a paper measuring tape and detailed instructions. High-risk waist size. An indicator for respondent waist sizes measured at age 30 being ≥88 cm for females and ≥102 cm for men. These waist sizes are considered to be strongly associated with subsequent risk of cardiometabolic diseases (Klein et al., 2007). Parental imprisonment. When children were ages 5 and 14, biological mothers were asked if they or their current partner had been detained in prison. From these variables, two measures were constructed for (1) if the mother or current partner had ever been detained in prison before age 5 and (2) if the mother reported she or current partner had ever been detained in prison at age 5 and/or age 14 interviews. We note that ‘current partner’ to the biological mother may be either the biological father or a non-biological father. Maternal education. Maternal education is based on the biological mother’s self-reported educational level prior to birth. Using this measure, we construct an indicator for if the mother had not completed secondary school or had completed any tertiary educational studies. Mother’s education is used to control for child socio-economic status. Child birth weight. Measured birth weight in grams from obstetric reports. Child birth weight is a significant predictor of increased BMI and cardiometabolic disease risk in adulthood (Jornayvaz et al., 2016; Kinge, 2017). Non-European ethnicity. Based on a constructed classification, an indicator for the child being of non-European ethnicity (i.e., of Asian and/or Indigenous Australian descent). This control allows us to adjust for potential ethnic variation in the sample. Sex. Respondent sex at birth was classified from obstetric data. Sex at birth is both a control and a potential moderator of results in the sample. Pregnancy status. At child ages 21 and 30, an indicator for female respondents who report being pregnant at the time of interview. To better compare results with Roettger and Boardman (2012), we include a control for whether or not the individual is pregnant to control for increased BMI related to pregnancy. We note that removing pregnant females from the sample did not substantively alter the results presented below. We used multivariate OLS regression to estimate continuous outcomes at ages 14, 21, and 30. We used multivariate logistic regression to estimate the odds of having a high-risk waist size at age 30. In addition, we used multivariate growth-curve models to estimate the association between parental imprisonment and time-varying measures of 1) BMI at ages 14, 21, and 30 and 2) systolic and DBP at ages 21 and 30. In doing so, we model the time-varying variations of these measures associated with parental imprisonment by using a random individual-level intercept. As Curran and colleagues note (Curran et al., 2010), this analysis allows us to make use of partial-data for individuals across waves and also determine if the associations observed at single waves hold as individuals progress through the life course. The estimation of change over time provides more robust findings holding across waves, relative to single-wave trends which may be influenced by single-wave attrition. To examine sex differences in risk, we estimate results for (1) pooled sex, (2) males only, and (3) females only. We conducted analyses using Stata 15.1. In the statistical analyses from regression analyses presented below, we report the unstandardized beta coefficient or adjusted odds ratio, along with 95% confidence intervals.