R package regmedint: extension of regression-based causal mediation analysis with effect measure modification by covariates

This is the supplementary document of R package regmedint that implements the extension of the regression-based causal mediation analysis first proposed by Valeri and VanderWeele (2013, 2015). It supports including effect measure modification by covariates (treatment-covariate and mediator-covariate product terms in mediator and outcome regression models), and also accommodates the original SAS macro and PROC CAUSALMED procedure in SAS when there is no effect measure modification.

The Efficacy of Low-dose Aspirin in Pregnancy Among Women in Malaria-endemic Countries

BackgroundLow dose aspirin (LDA) is an effective strategy to reduce preterm birth. However, LDA might have differential effects globally, based on the etiology of preterm birth. In some regions, malaria in pregnancy could be an important modifier of LDA on birth outcomes and anemia. MethodsThis is a sub-study of the ASPIRIN trial, a multi-national, randomized, placebo controlled trial evaluating LDA effect on preterm birth. We enrolled a convenience sample of women in the ASPIRIN trial from the Democratic Republic of Congo (DRC), Kenya and Zambia. We used quantitative polymerase chain reaction to detect malaria. We calculated crude prevalence proportion ratios (PRs) for LDA by malaria for outcomes, and regression modelling to evaluate effect measure modification. We evaluated hemoglobin in late pregnancy based on malaria infection in early pregnancy.Results1,446 women were analyzed, with a malaria prevalence of 63% in the DRC site, 38% in the Kenya site, and 6% in the Zambia site. Preterm birth occurred in 83 (LDA) and 90 (placebo) women, (PR 0.92, 95% CI 0.70, 1.22), without interaction between LDA and malaria (p=0.75). Perinatal mortality occurred in 41 (LDA) and 43 (placebo) pregnancies, (PR 0.95, 95% CI 0.63, 1.44), with an interaction between malaria and LDA (p=0.014). Hemoglobin was similar by malaria and LDA status.ConclusionsMalaria in early pregnancy did not modify the effects of LDA on preterm birth, but modified the effect of LDA on perinatal mortality. This effect measure modification deserves continued study as LDA is used in malaria endemic regions.

511Effects of educational activities prior to school entry are more important for socioeconomically disadvantaged children

Background We examined whether time in educational activities at 2-3 years and language ability at school entry differed among children from diverse socioeconomic backgrounds. Methods Data were drawn from the Longitudinal Study of Australian Children (n = 4253). Educational activities were collected using time-use diaries. Household income was measured using parent self-report. Language ability was assessed using the Peabody Picture Vocabulary Test. Marginal structural models with inverse probability of treatment weights were used to assess modification of the effect of time spent in educational activities on language ability by income. Results Children exposed to both <30 minutes/day in educational activities and being in a low-income household were at greater risk of poorer outcomes than the simple sum of their independent effects. Compared with children who spent ≥30 minutes/day in educational activities and from higher-income households, children who experienced <30 minutes/day from low-income households had a 2.30 (95% CI: 1.88, 2.80) higher risk of having lower language ability. The Relative Excess Risk Due to Interaction of 0.15 (95% CI:-0.15, 0.67) was greater than zero indicating a super-additive effect measure modification by income. Conclusions Our results suggest that boosting time in educational activities to 30 minutes/day would benefit the lower-income group relatively more than the higher-income group. Key messages If there was an intervention of sufficient dose to increase the amount of time spent in educational activities to 30 minutes/day for children in the lower-income group, the risk of children having lower language ability at school entry could be reduced by 45%.

1198The gendered associations between poverty, marital status, and mental health in older Australians

Background Compared to men, older women have poorer mental health and are more vulnerable to poverty. Few studies have examined how gender, marital status and poverty are inter-related and are associated with mental health among this group. Methods Drawing on 17 waves of the Housing Income and Labour Dynamics in Australia survey, fixed-effects regression analysis was utilised to examine associations between: (1) relative poverty (<50% median household income) and mental health (MHI-5); (2) marital status and poverty, among Australians aged 65+years. Effect measure modification (EMM) of the association between relative poverty and mental health by marital status was also assessed. Results Relative poverty was associated with poorer mental health among women (−0.62, 95%CI−1.06,−0.18), but not men. Being divorced/separated was associated with increased odds of relative poverty for women (OR 3.93, 95%CI 2.54,6.06) but not men, and widowhood was associated with relative poverty in women (OR 4.23, 95%CI 3.37,5.29), and men (OR 2.22, 95%CI 1.62,3.05). There was no evidence of EMM of the relationship between relative poverty and mental health by marital status for either gender. Conclusions This study provides evidence that relative poverty is a major determinant of mental health in older Australian women. Addressing gender inequities in lifetime savings, as well as in division of acquired wealth post marital loss, may mitigate these disparities. Key messages A gendered double burden of mental health issues and poverty exists for older Australian women, requiring urgent policy attention.

Built Environment and HIV Linkage to Care in Rural South Africa

Background We assessed built environment (residential density, landuse mix and aesthetics) and HIV linkage to care (LTC) among 1,681 (18–49 years-old) residents of 15 Mpumalanga villages, South Africa. Methods Multilevel models (linear-binomial) were used for the association between built environment, measured using NEWS for Africa, and LTC from a clinical database of 9 facilities (2015–2018). Additionally, we assessed effect-measure modification by universal test-and-treat policy (UTT). Results We observed, a significant association in the adjusted 3-month probability of LTC for residential density (risk difference (RD)%: 5.6, 95%CI: 1.2–10.1), however, no association for land-use mix (RD%: 2.4, 95%CI: −0.4, 5.2) and aesthetics (RD%: −1.2, 95%CI: −4.5–2.2). Among those diagnosed after UTT, residents of high land-use villages were more likely to link-to-care than those of low land-use villages at 12 months (RD%: 4.6, 95%CI: 1.1–8.1, p < 0.04), however, not at 3 months (RD%: 3.0, 95%CI: −2.1–8.0, p > 0.10). Conclusion Findings suggest, better built environment conditions (adequate infrastructure, proximity to services etc.) help facilitate LTC. Moreover, UTT appears to have a protective effect on LTC.

Hypertension, BMI, and cardiovascular and cerebrovascular diseases

Hypertension is associated with body mass index (BMI) and cardiovascular and cerebrovascular diseases (CCDs). Whether hypertension modifies the relationship between BMI and CCDs is still unclear. We examined the association between BMI and CCDs and tested whether effect measure modification was present by hypertension. We identified a population-based sample of 3,942 participants in Shuncheng, Fushun, Liaoning, China. Hypertension was defined as any past use of antihypertensive medication or having a measured systolic/diastolic blood pressure ≥130/80 mm Hg. BMI was calculated from measured body weight and body height. Data on diagnosed CCDs were self-reported and validated in the medical records. We used logistic regression models to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for associations between BMI and CCDs. Higher BMI was associated with increased odds of having CCDs (OR = 1.19, 95% CI: 1.07–1.31). This association was significantly modified by hypertension (P for interaction <0.001), with positive associations observed among hypertensive individuals (OR = 1.28, 95% CI: 1.14–1.42). Age, sex, and diabetic status did not modify the relationship between BMI and CCDs (all P for interaction >0.10). Although higher BMI was associated with increased odds of CCDs, the relationship was mainly limited to hypertensive patients.

Does disability modify the relationship between labour force status and psychological distress among young people?

ObjectiveTo examine the association between labour force status, including young people who were unemployed and having problems looking for work, and psychological distress one year later. We then assessed whether this association is modified by disability status.MethodsWe used three waves of cohort data from the Longitudinal Surveys of Australian Youth. We fitted logistic regression models to account for confounders of the relationship between labour force status (employed, not in the labour force, unemployed and having problems looking for work) at age 21 years and psychological distress at age 22 years. We then estimated whether this association was modified by disability status at age 21 years.ResultsBeing unemployed and having problems looking for work at age 21 years was associated with odds of psychological distress that were 2.48 (95% CI 1.95 to 3.14) times higher than employment. There was little evidence for additive effect measure modification of this association by disability status (2.52, 95% CI −1.21 to 6.25).ConclusionsYoung people who were unemployed and having problems looking for work had increased odds of poor mental health. Interventions should focus on addressing the difficulties young people report when looking for work, with a particular focus on supporting those young people facing additional barriers to employment such as young people with disabilities.

Effect Measure Modification of Multimorbidity On the Association Between Regularity of General Practitioner Contact and Potentially Avoidable Hospitalisations

IntroductionRegular contact with a general practitioner (GP) has been shown to lower the risk of potentially avoidable hospitalisations (PAHs) independently of continuity of provider and frequency of contact. Multimorbidity affects between 55 and 98% people aged 65+ years and continues to place pressure on healthcare systems globally. However, little is known about its impact on the relationship between continuity of primary care and PAHs. Objectives and ApproachA retrospective, longitudinal cohort study using survey data linked to routinely-collected administrative health data from the 45 and Up Study conducted in New South Wales, Australia was used to investigate the effect measure modification by multimorbidity on the relationship between regularity of GP contact and PAHs. Multimorbidity was assessed using the Rx-Risk comorbidity score, which captures the number of condition groups, assigned based on medicine dispensing records, using a 5-year look-back period. PAHs were: (i) any unplanned hospitalisations, (ii) chronic ambulatory care sensitive conditions (ACSC) hospitalisations or (iii) unplanned ACSC hospitalisations. Multivariable logistic regression and population attributable fractions (PAF) were used to examine effect measure modification by multimorbidity. ResultsHigher GP regularity was significantly associated with a reduction in the probability of each PAH type. This reduction diminished with increasing multimorbidity with the effect measure modification most apparent for chronic ACSC and unplanned chronic ACSC hospitalisations. The PAF of moving to the highest quintile of regularity significantly reduced with increasing multimorbidity. For example, a reduction in the PAF of unplanned ACSC hospitalisations of 31.1% was observed in those with a RX Risk score of >10 (17.8%) compared with those with no multimorbidity (48.9%). Conclusion / ImplicationsWeakening of the relationship between GP visit regularity and PAHs with increasing levels of multimorbidity suggests a need to focus on improving primary care support to prevent PAHs for patients with multimorbidity.