Exploring the utility of synthetic data to extract more value from sensitive health data assets: A focused example in perinatal epidemiology.

BACKGROUND: Privacy, access and security concerns can hinder the availability of health data for research. The use of synthesised data in place of de-identified electronic health records (EHRs) presents an opportunity to conduct research while minimising privacy concerns. OBJECTIVES: To examine whether synthesised data can replicate two prenatal epidemiological associations: between prenatal smoking and lower birthweight, and between prenatal mood disorders and lower birthweight, using data synthesised from de-identified health administrative data collections. METHODS: We generated two synthetic datasets, using parametric and non-parametric data generating methods, and examined the synthetic data for evidence of privacy concerns. Next, univariable and multivariable logistic regression was utilised to estimate the associations in both synthetic datasets, with results then compared to the real data. RESULTS: Both synthesised datasets performed well in identifying the reduction in birthweight associated with prenatal smoking, while the non-parametric data underestimated the reduction in birthweight associated with prenatal mood disorders. Improbable relationships between some variables were identified in the parametric synthesised data, however, these can be addressed with simple rules during data synthesis. No duplicate rows (i.e., exact copies of de-identified data) were found in the parametric data, while only 0.6% of the rows in the non-parametric data were duplicated. CONCLUSIONS: Both synthesised datasets performed well in replicating the statistical properties of the original data while addressing privacy issues. Data synthesis methods provide an opportunity for researchers to utilise health data while managing privacy and security concerns.

Duration of opioid use and association with socioeconomic status, daily dose and formulation: a two-decade population study in Queensland, Australia.

Background There is an association between the duration of prescription opioids use and an increased risk of serious harm, often unintentional. Objective (1) Describe the trends in duration of prescription opioids dispensing and, (2) determine the risk of long-term use (≥4 months) based on patients' socioeconomic status, daily dose in oral daily morphine milligram equivalent, and opioid formulation. Setting Residents of Queensland (2,827,727), Australia from the age 18 years and who were dispensed pharmaceutical opioids from 1 January 1997 to 31 December 2018. Method Retrospective, longitudinal population-based analysis using data obtained from the Monitoring of Drugs of Dependence system of the Monitored Medicines Unit of Queensland Health. Main outcome measure Contribution of socioeconomic status, and daily dose and opioid formulation (modified-release or immediate-release) to the risk of long-term opioid use. Results There was little difference between the number of patients dispensed opioids for ≥4 months and ≤3 months between 1997 and 2011. Thereafter, the number for those using opioids long-term increased. The highest risk of having opioids dispensed for ≥4 months were for patients in the lowest level of socioeconomic status (adjusted odds ratio 1.36; 95% CI, 1.34, 1.38), compared to people in the highest socioeconomic status areas, followed by the low-socioeconomic status areas, mid-socioeconomic status areas, and high-socioeconomic status areas respectively. The risk of being dispensed prescription opioids for ≥4 months significantly increased as the dose increased: adjusted odds ratio 1.73; 95% CI, 1.71, 1.75, adjusted odds ratio 1.89; 95% CI, 1.87, 1.92, and adjusted odds ratio 3.63; 95% CI, 3.58, 3.69 for the ≥20 to <50 oral daily morphine milligram equivalent, ≥50 to <100 oral daily morphine milligram equivalent and ≥100 oral daily morphine milligram equivalent dose categories, respectively. Conclusion Higher doses and living in a low socioeconomic status areas were associated with increased risk of long-term dispensing of opioid prescriptions.

Pharmaceutical opioids utilisation by dose, formulation, and socioeconomic status in Queensland, Australia: a population study over 22 years.

Background Prescription opioids are a central aspect of pain management and as the prevalence of pain is increasing so is the rate of use of prescription opioids. Increased opioid prescriptions increases the risk of deaths and morbidity. Objective To (a) describe the 22-year trend of prescription opioid dispensing in Queensland, (b) examine the effect of opioid dose, formulation and socioeconomic status on the number of prescriptions dispensed. Design/setting Retrospective analysis of data from the Monitoring of Drugs of Dependence system of the Monitored Medicines Unit of Queensland Health, Australia. Participants Queensland residents (3.3 million) from 18 years old dispensed 18.8 million opioid prescriptions from January 1997 to December 2018. Results Opioid prescriptions dispensed annually increased to over two million in 2018 from about 150,000 prescriptions in 1997. The number of prescriptions for modified-release formulations dispensed annually was three times higher compared to the immediate-release formulations. Oxycodone accounted for over 60% of prescriptions for pharmaceutical opioids since 2013. There was an increase in the number of prescriptions dispensed as socioeconomic status decreased and modified-release opioid formulations positively affects the pattern of dispensing. The highest increase in number of prescriptions dispensed (for all opioids) was observed among the high socioeconomic status (IRR = 1.25, 95% CI 1.25, 1.26). The disparities in the annual number of prescriptions across dose categories are wider in the modified-release than the immediate-release formulations. Conclusion The dispensing of opioids increased significantly in Queensland. There was a positive relationship between the increased dispensing of opioids and locations of lower socioeconomic status.

Factors influencing repeated teenage pregnancy: a review and meta-analysis.

OBJECTIVE: Existing evidence of predictors of repeated teenage pregnancy has not been assessed rigorously. This systematic review provides a comprehensive evaluation of protective and risk factors that are associated with repeated teenage pregnancy through a metaanalytical consensus. DATA SOURCES: We used PubMed, EMBASE, CINAHL, ProQuest, PsychINFO, ScienceDirect, Scopus, and Web of Science databases from 1997-2015 and the reference list of other relevant research papers and related reviews. STUDY ELIGIBILITY CRITERIA: Eligibility criteria included (1) epidemiologic studies that analyzed factors associated with repeated pregnancy or birth among adolescents <20 years of age who were nulliparous or experienced at least 1 pregnancy, and (2) experimental studies with an observational component that was adjusted for the intervention. STUDY APPRAISAL AND SYNTHESIS METHODS: We performed narrative synthesis of study characteristics, participant characteristics, study results, and quality assessment. We also conducted random-effects and quality-effects metaanalyses with meta-regression to obtain pooled odds ratios of identified factors and to determine sources of between-study heterogeneity. RESULTS: Twenty-six eligible epidemiologic studies, most from the United States (n=24), showed >47 factors with no evidence of publication bias for each metaanalysis. Use of contraception (pooled odds ratio, 0.60; 95% confidence interval, 0.35-1.02), particularly long-acting reversible contraceptives (pooled odds ratio, 0.19; 95% confidence interval, 0.08-0.45), considerably reduced repeated teenage pregnancy risk. Among studies about contraception, the number of follow-up visits (adjusted coefficient, 0.72; P=.102) and country of study (unadjusted coefficient, 2.57; permuted P=.071) explained between-study heterogeneity. Education-related factors, which included higher level of education (pooled odds ratio, 0.74; 95% confidence interval, 0.60-0.91) and school continuation (pooled odds ratio, 0.53; 95% confidence interval, 0.33-0.84), were found to be protective. Conversely, depression (pooled odds ratio, 1.46; 95% confidence interval, 1.14-1.87), history of abortion (pooled odds ratio, 1.66; 95% confidence interval, 1.08-2.54), and relationship factors, such as partner support, increased the repeated teenage pregnancy risk. CONCLUSION: Contraceptive use, educational factors, depression, and a history of abortion are the highly influential predictors of repeated teenage pregnancy. However, there is a lack of epidemiologic studies in low- and middle-income countries to measure the extent and characteristics of repeated teenage pregnancy across more varied settings.

Low-intensity case management increases contact with primary care in recently released prisoners: a single-blinded, multisite, randomised controlled trial.

BACKGROUND: The world prison population is large and growing. Poor health outcomes after release from prison are common, but few programmes to improve health outcomes for ex-prisoners have been rigorously evaluated. The aim of this study was to evaluate the impact of individualised case management on contact with health services during the first 6 months post-release. METHODS: Single-blinded, randomised, controlled trial. Baseline assessment with N=1325 adult prisoners in Queensland, Australia, within 6 weeks of expected release; follow-up interviews 1, 3 and 6 months post-release. The intervention consisted of provision of a personalised booklet ('Passport') at the time of release, plus up to four brief telephone contacts in the first 4 weeks post-release. RESULTS: Of 1179 eligible participants, 1003 (85%) completed ≥1 follow-up interview. In intention-to-treat analyses, 53% of the intervention group and 41% of the control group reported contacting a general practitioner (GP) at 1 month post-release (difference=12%, 95% CI 5% to 19%). Similar effects were observed for GP contact at 3 months (difference=9%, 95% CI 2% to 16%) and 6 months (difference=8%, 95% CI 1% to 15%), and for mental health (MH) service contact at 6 months post release (difference=8%, 95% CI 3% to 14%). CONCLUSIONS: Individualised case management in the month after release from prison increases usage of primary care and MH services in adult ex-prisoners for at least 6 months post-release. Given the poor health profile of ex-prisoners, there remains an urgent need to develop and rigorously evaluate interventions to increase health service contact in this profoundly marginalised population. TRIAL REGISTRATION NUMBER: ACTRN12608000232336.

Cohort Profile Update: The Mater-University of Queensland Study of Pregnancy (MUSP).

The Mater-University of Queensland Study of Pregnancy (MUSP) and its outcomes began in 1981 with data collected on 7223 pregnant woman-child pairs (6753 mothers, of whom 520 had 2 study children, less 50 who had multiple births). These women, and their children, were initially followed for up to 21 years. Since then there have been additional follow-ups of the mothers (27 years) and their children (30 years). There has also been a substantial increase in the breadth of topics addressed, with the collection of biological samples, the administration of structured clinical assessments of mental health and cognitive capacity, and markers of physical health such as lung function and blood pressure. MUSP was originally developed as a birth cohort study. It has become a longitudinal study of growth, development and ageing with an emphasis on the generational transmission of a wide range of factors impacting on adult health outcomes. We welcome interest in our study; for study background and publications visit [www.socialscience.uq.edu.au/musp] or contact [j.najman@uq.edu.au].

Does early socio-economic disadvantage predict comorbid alcohol and mental health disorders?

BACKGROUND: Alcohol and mental health disorders are highly prevalent in the general population, with co-occurrence recognised as a major public health issue. Socio-economic factors are frequently associated with both disorders but their temporal association is unclear. This paper examines the association between prenatal socio-economic disadvantage and comorbid alcohol and mental health disorders at young adulthood. METHODS: An unselected cohort of women was enrolled during early pregnancy in the large longitudinal Mater-University of Queensland Study of Pregnancy (MUSP), at the Mater Misericordiae Public Hospital in Brisbane, Australia. The mothers and their offspring were followed over a 21 year period. Offspring from the MUSP birth cohort who provided full psychiatric information at age 21 and whose mothers provided socioeconomic information at baseline were included (n=2399). Participants were grouped into no-disorder, mental health disorder only, alcohol disorder only or comorbid alcohol and mental health disorders according to DSM-IV diagnoses at age 21 as assessed by the Composite International Diagnostic Interview. We used multivariate logistic regression analysis to compare associations of disorder group with single measures of prenatal socio-economic disadvantage including family income, parental education and employment, and then created a cumulative scale of socioeconomic disadvantage. RESULTS: Greater socio-economic disadvantage was more strongly associated with comorbidity (OR 3.36; CI95 1.37, 8.24) than with single disorders. This relationship was not fully accounted for by maternal mental health, smoking and drinking during pregnancy. CONCLUSION: Multiple domains of socio-economic disadvantage in early life are associated with comorbid alcohol and mental health disorders.

Poor health and social outcomes for ex-prisoners with a history of mental disorder: a longitudinal study.

OBJECTIVE: To examine the association between self-reported lifetime diagnosis of mental disorder and health-related outcomes in prisoners during the first six months after release. METHODS: We interviewed 1,324 adult prisoners in Queensland, Australia, within six weeks of expected release and one, three and six months post-release. Outcomes of interest included health service access, housing, employment, substance use and criminal activity. We used multivariate logistic regression to investigate the association between self-reported, lifetime diagnosis of mental disorder and these health-related outcomes post-release, adjusting for pre-existing disadvantage. RESULTS: 43.4% of participants reported a lifetime diagnosis of mental disorder. This group had increased crude odds of poor outcomes across all evaluated domains. After adjusting for pre-existing disadvantage, significantly increased odds of poor outcomes persisted in the substance use, mental health, crime and health service access domains. CONCLUSIONS: People with a history of mental disorder experience particularly poor outcomes following release from prison that are not fully explained by pre-existing disadvantage. IMPLICATIONS: Evidence-based transitional programs for prisoners with a history of mental disorder should be provided at a level commensurate with need.

Randomised controlled trial of a service brokerage intervention for ex-prisoners in Australia.

BACKGROUND: Health outcomes after release from prison are typically poor with elevated rates of morbidity and mortality widely documented. Under-utilisation of health services contributes to these outcomes, but interventions to increase health service utilisation in ex-prisoners are in their infancy and few have been rigorously evaluated. METHODS: Single-blinded randomised controlled trial of a service brokerage intervention (the 'Passports study') for N = 1325 adult ex-prisoners in Queensland, Australia. Participants in the intervention group received a personalised booklet summarising their health status and identifying appropriate community health services; trained workers made weekly telephone contact in the first 4 weeks post-release to identify health needs and facilitate health service utilisation. Participants in the control arm received usual care. Baseline data were collected within 6 weeks of expected release from custody with follow-up telephone interviews 1, 3 and 6 months post-release. Participant identities were linked with federal health service utilisation records, a national death register and corrective services records, two years post-release. The primary outcome was self-reported health service utilisation in the first 6 months post-release. RESULTS: Between 2008 and 2010 1976 prisoners were screened for eligibility, 1665 met eligibility criteria and 1325 were recruited; 665 were randomised to the intervention and 660 to the control condition. Participants were broadly representative of adults being released from prison in Queensland except that women were intentionally oversampled (21% vs. 11%). CONCLUSIONS: Outcomes from this large RCT will provide the first robust evidence of the effect of service brokerage on health service utilisation and health outcomes for ex-prisoners.

Monte Carlo simulation of the cost-effectiveness of sample size maintenance programs revealed the need to consider substitution sampling.

OBJECTIVE: To assess the cost-effectiveness of sample size maintenance programs in a prospective cohort. STUDY DESIGN AND SETTING: The Living with Diabetes Study in Queensland, Australia is a longitudinal survey providing a comprehensive examination of health care utilization and disease progression among people with diabetes. Data from this study were used to compare the cost-effectiveness of a program incorporating substitution sampling with two alternative programs: "no follow-up" and "usual practice." RESULTS: A program involving substitution sampling was shown to be the most effective with an additional 3,556 complete responses (compared with a "no follow-up" program) and an additional 2,099 complete responses (compared with "usual practice"). An incremental analysis through a Monte Carlo simulation found substitution sampling to be the most cost-effective option for maintaining sample size with an incremental cost-effective ratio of $54.87 (95% uncertainty interval $52.68-$57.25) compared with $87.58 ($77.89-$100.09) for "usual practice." CONCLUSIONS: Based on the available data, a program involving substitution sampling is economically justified and should be considered in any approach with the aim of maintaining sample size. There is, however, a continuing need to evaluate the effectiveness of this option on other outcome measures, such as bias.

The self-reported personal wellbeing of a sample of Australian injecting drug users.

AIMS: To examine the self-reported personal wellbeing of a sample of Australian injecting drug users (IDU) using a standardized instrument and determine the key correlates of variations in self-reported personal wellbeing. DESIGN, SETTING AND PARTICIPANTS: Cross-sectional survey of 881 Australian IDU. MEASUREMENTS: Self-reported personal wellbeing collected using the Personal Wellbeing Index (PWI). FINDINGS: IDU scored significantly lower than the general Australian population on the PWI and all subscales. Lower PWI scores were associated with a range of socio-demographic, drug use and other health and social characteristics. Across all PWI subscales, lower personal wellbeing scores were associated with unemployment, past 6-month mental health problems and more frequent injecting (all P < 0.05). CONCLUSIONS: The PWI is sufficiently sensitive to distinguish between IDU and the general population, and to identify key correlates of PWI among IDU. Some domains canvassed within the scale, such as health, standard of living and life achievements, are well within the scope of current intervention strategies, such as pharmacotherapy maintenance treatment and housing and employment support services. This suggests that the PWI could be useful in clinical settings by allowing structured identification of the areas of a person's life to be addressed as a part of a treatment regimen. In order to inform targeted prevention and intervention efforts, longitudinal studies of PWI and its correlates among IDU are required.

Common symptoms during pregnancy to predict depression and health status 14 years post partum.

OBJECTIVE: To examine the prospective association between symptoms commonly experienced during pregnancy and the mental and general health status of women 14 years post partum. METHODS: Data used were from the Mater-University of Queensland Study of Pregnancy, a community-based prospective birth cohort study begun in Brisbane, Australia, in 1981. Logistic regression analyses were conducted. RESULTS: Data were available for 5118 women. Women who experienced a higher burden of symptoms during pregnancy were at greater risk of becoming depressed and reporting poorer health status 14 years post partum. Women who experienced major problems during pregnancy were 4 times more likely to be depressed and nearly 8 times more likely to report poorer health status 14 years after the index pregnancy compared with women who experienced few problems. CONCLUSIONS: Findings suggest that pregnant women who experience common symptoms during pregnancy are likely to experience poorer mental and self-reported general health 14 years after the pregnancy.

Prediction of adolescent smoking from family and social risk factors at 5 years, and maternal smoking in pregnancy and at 5 and 14 years.

AIMS: This study examines associations between maternal smoking and family, social or child risk factors when the child is aged 5 and adolescent smoking. The influence of mothers who smoke in pregnancy or continue to smoke at 14 years was also examined. DESIGN: The Mater-University of Queensland Study of Pregnancy is a prospective cohort study. PARTICIPANTS: Participants included 8556 women enrolled between 1981 and 1984 at their first antenatal visit. Completed questionnaires were obtained for 7223 offspring, comprising the study birth cohort. Of the 7223 eligible children a total of 4541 had information on both maternal smoking when the child was aged 5 years and adolescent smoking at 14 years. MEASUREMENTS: Measures included maternal smoking during pregnancy and when the child was aged 5 and 14 years, child smoking at 14 years, maternal alcohol use, child behaviour problems and social and demographic variables. FINDINGS: Adolescent smoking was predicted by a risk score at 5 years involving maternal smoking and alcohol use, non-married status, having a partner who had ever been arrested, having four or more children in the household, and child aggression at 5 years. Continued maternal smoking from 5 to 14 years was associated strongly with adolescent smoking. There was also evidence that smoking in late pregnancy may exert an independent effect on adolescent smoking. CONCLUSIONS: The results suggest the possibility of a direct effect of prenatal smoking on adolescent smoking and highlight a set of environmental risk factors in the development of adolescent smoking. These risk factors may be used as early warning signs that intervention may be needed, and given the similarities with risk factors for other adverse childhood outcomes, the benefits of early intervention may extend beyond smoking to other problem behaviours. The possibility of being able to predict other disorders, because of these associations, also warrants further investigation.

Changes in mental health status amongst children of migrants to Australia: a longitudinal study.

This paper examines the mental wellbeing of children of Australian migrants. Migration can be viewed as a natural experiment in which persons of one culture have their beliefs, values and behaviours challenged by the host culture. Such a process could be expected to lead to impaired mental health amongst migrants and their children. This paper investigates the relationship between migrants' region of origin, length of stay in Australia and indicators of impaired mental health. The data were taken from the Mater-University of Queensland Study of Pregnancy (MUSP), a longitudinal study of mothers and children which started in Brisbane, Australia, in 1981. The study comprises a cohort of over 5,000 women interviewed at their first ante-natal clinic visit and followed up at 3-5 days, six months, 5 and 14 years after the baby was born. Results showed no significant differences between the mental health of 'second generation' children and their Australian counterparts. Length of stay in Australia was not associated with internalising symptoms (anxiety and depression). There was a positive association, however, between the length of stay in Australia and increased externalising problems (aggression and delinquency) amongst the children at both 5- and 14-years follow-up. We conclude that children of migrants do not differ from comparable children of Australian-born parents in their mental health. Children of migrant parents have fewer symptoms of some behaviour problems in the years after their arrival in Australia, but these rates increase to the Australian level over time.