Australian Longitudinal Study of Women’s Health (ALSWH) Analysis Code
R and Stata Analysis Code
This repository contains R code used in articles using the ALSWH.
Physical activity across midlife and health-related quality of life in Australian women: A target trial emulation using a longitudinal cohort
Code for all analysis in the article examining the causal effects of meeting physical activity on health-related quality of life (Nguyen-duy et al 2024) published in Plos Medicine: https://doi.org/10.1371/journal.pmed.1004384.
| Description | R Code |
|---|---|
| 1 - Data extraction - pull relevant variables from each wave | Data extraction |
| 2 - Merge data - merge waves and create derived variables | Merge data |
| 3 - Multiple imputation - impute intermittent missing data | Imputation |
| 4 - Final data creation - finalise imputed data and structure for analysis | Finalise data |
| 5 - LTMLE analysis - using dynamic regimes based on age, using the package ‘ltmle’ (1). | LTMLE analysis |
| 6 - Sensitivty analysis using lower physical activity cut-point. | Sensitivity 1 |
| 7 - Sensitivty analysis excluding variables wholly missing in some waves. | Sensitivity 1 |
| 8 - Pool results across imputations and create analysis figures | Pool results |
| 9 - Create plots of results using ggplot | Create plots |
| 10 - Generate ‘table 1’ of baseline descriptive statistics | Descriptives |
| 11 - E-value analysis to test sensitivity to unmeasured confounding | E-value analysis |
| 12 - Create summary of missing data | Missing data |
Causal effects of physical activity on mortality
Code for all analysis in the article examining the causal effects of meeting physical activity on mortality (Nguyen-duy et al 2025, under review).
| Description | R Code |
|---|---|
| 1 - Data extraction - pull relevant variables from each wave | Data extraction |
| 2 - Merge data - merge waves and create derived variables | Merge data |
| 3 - Multiple imputation - impute intermittent missing data | Imputation |
| 4 - Final data creation - finalise imputed data and structure for analysis | Finalise data |
| 5 - Analysis of all-cause mortality - using dynamic regimes based on age, using the package ‘ltmle’ (1). | All-cause analysis |
| 6 - Analysis of CVD mortality - using dynamic regimes based on age, using the package ‘ltmle’ (1). | CVD analysis |
| 7 - Analysis of Cancer mortality - using dynamic regimes based on age, using the package ‘ltmle’ (1). | Cancer analysis |
| 8 - Pool results across imputations and create analysis figures | Pool results |
| 9 - Create plots to graphically report the analysis findings | Create plots |
Causal effects of loneliness on all-cause mortality
Code for all analysis in the article examining the causal effects of loneliness on mortality (HaGani et al 2025, under review).
| Description | R Code |
|---|---|
| 1 - Data extraction - pull relevant variables from each wave | Data extraction |
| 2 - Merge data - merge waves and create derived variables | Merge data |
| 3 - Multiple imputation - impute intermittent missing data | Imputation |
| 4 - Final data creation - finalise imputed data and structure for analysis | Finalise data |
| 5 - Analysis of all-cause mortality - using dynamic regimes based on age, using the package ‘ltmle’ (1). | All-cause analysis |
| 6 - Post-hoc sensitivity analysis adjusting for baseline conditions rather than excluding. | Sensitivity |
| 7 - Pool results across imputations and create analysis figures | Pool results |
| 8 - Create plots to graphically report the analysis findings | Create plots |
| 9 - E-Value analysis of unmeasured confounding | EValue analysis |
| 10 - Missing data summary for appendix | Missing data |
| 11 - Descriptive statistics on unadjusted mortality incidence. | Mortality descriptives |
| 12 - Socio-demographic descriptives for Table 1. | Sociodemographic descriptives |
Physical activity and incident obesity: causal inference analysis in Australian women aged 45 years and older
Code for all analysis in the article examining the causal effects of physical activity on incident obesity (Tarp et al 2025), in progress.
| Description | R Code |
|---|---|
| 1 - Data extraction - pull relevant variables from each wave | Data extraction |
| 2 - Merge data - merge waves and create derived variables | Merge data |
| 3a - Multiple imputation - impute intermittent missing data | Imputation |
| 3b - Multiple imputation for sensitivity analysis | Imputation |
| 4 - Final data creation - finalise imputed data and structure for analysis | Finalise data |
| 5a - Linearity tests for functional form of PA in primary analysis | Primary linearity |
| 5b - Linearity tests for functional form of PA in categorical analysis | Categorical linearity |
| 5c - Linearity tests for functional form of PA in analysis of ‘severe’ obesity | Severe linearity |
| 5d - Linearity tests for functional form of PA in analysis of five-percent weight gain | Five-percent linearity |
| 5e - Linearity tests for functional form of PA in analysis of ten-percent weight gain | Ten-percent linearity |
| 5f - Combine and summarise linearity tests | Combine linearity |
| 6a - Primary analysis | Primary analysis |
| 6b - Categorical analysis | Categorical analysis |
| 6c - Analysis of ‘severe’ obesity | Severe analysis |
| 6d - Analysis of five-percent weight gain | Five-percent analysis |
| 6e - Analysis of ten-percent weight gain | Ten-percent analysis |
| 6f - Secondary analysis stratified by level of education | Stratified analysis |
| 6g - Sensitivity analysis controlling for descendants of possible unmeasured confounder | Sensitivity analysis 1 |
| 6h - Sensitivity analysis excluding variables that were wholly imputed in some waves | Sensitivity analysis 2 |
| 7 - Pool results across imputations and create analysis figures | Pool results |
| 8 - Create plots to graphically report the analysis findings | Create plots |
| 9 - Socio-demographic descriptives for Table 1 and on unadjusted incidence of outcomes. | Sociodemographic descriptives |
| 10 - E-Value analysis of unmeasured confounding | EValue analysis |
| 11 - Missing data summary for appendix | Missing data |
- Lendle SD, Schwab J, Petersen ML, van der Laan MJ. ltmle: An R Package Implementing Targeted Minimum Loss-Based Estimation for Longitudinal Data. Journal of Statistical Software. 2017;81(1):1-21.
- Vermunt JK. Latent Class Modeling with Covariates: Two Improved Three-Step Approaches. Political Analysis. 2010;18(4):450-469.
Dr. Philip J Clare, PhD
Biostatistician at the Prevention Research Collaboration, University of Sydney.