Dataset DOI: 10.5061/dryad.sn02v6xm9

Description of the data and file structure

Exploratory factor analysis was used to derive a SDOH factor based on income, education level, and neighborhood area deprivation index (ADI) in Black and White participant race strata of the Atherosclerosis Risk in Communities (ARIC) study. Proteins associated with the SDOH factor at Bonferroni significance after adjusting for age and sex were then tested for their association with incident HF at a false discovery rate <0.05 using Cox regression adjusting for HF risk factors. Proteins associated with HF in both race strata were evaluated for potential mediation using causal mediation analysis, and for causal effects on HF using Mendelian randomization (MR) and colocalization. This analysis was replicated using the publicly available UK Biobank cohort data. 

Dataset access

ARIC data are not publicly available due to participant privacy and consent limitations, but are available to qualified investigators through the NHLBI BioLINCC repository (https://biolincc.nhlbi.nih.gov/) or by application to the ARIC Coordinating Center, subject to institutional review and data use agreements. Derived variables and analytic datasets generated during the current study are available from the corresponding author (Amil.Shah@utsouthwestern.edu) or coordinating center (aricdata@unc.edu) following DMDA form completion for ARIC approval.

Files and variables

File: Code_Repository.zip

Description: 

MAJOR FOLDERS: 

1. Source Functions
2. UKB Replication
3. Primary Analysis
4. Mendelian Randomization

Description of Major Folders: 

Source Functions:

• MediationWithInteraction.R: Code used for Causal Mediation analysis 
• load-data.R: Code used for loading datasets
• SourceFunctionsForPlotting.R: Plotting source code
• Plotting_Functions.R: Plotting source code
• make-tables.R: Tables source code
• MRSourceFunctions.R: Source functions for Mendelian Randomization analysis 
• SourceFilesUnivariateAnalysis.R: Source file for univariate linear regression
• temp-fig1.R: Figure source code
• NewFunctionsToSource.R: Source code for additional functions 

UKB Replication:

• UKB Replication.Rmd: Source code containing factor analysis, univariate linear regression and mediation analysis for the UK Biobank replication analysis 

Primary Analysis

• SDOH_Factor_Primary_Statistical_Analysis.Rmd (Primary statistical analysis- exploratory factor analysis using income, education, and area deprivation index (ADI) from ARIC to derive a SDOH-based latent factor group)

Mendelian Randomization/Colocalization

• SDOH Colocalization.Rmd (Colocalization analysis portion of MR)

Mendelian Randomization/Running MR/

• SDOH_MR_LVEDV_AGES_Pt1.R (MR analysis; Outcomes: LVEDV; GWAS: AGES)
• SDOH_MR_LVEF_AGES_Pt1.R (MR analysis; Outcomes: LVEF; GWAS: AGES)
• SDOH_MR_LVESV_Fenland_Pt1.R (MR analysis; Outcomes: LVESV; GWAS: Fenland)
• SDOH_MR_LVESV_Science_Pt1.R (MR analysis; Outcomes: LVESV; GWAS: Science)
• SDOH_MR_LVEF_Fenland_Pt1.R (MR analysis; Outcomes: LVEF; GWAS: Fenland)
• SDOH_MR_LVEF_Science_Pt1.R (MR analysis; Outcomes: LVEF; GWAS: Science)
• SDOH_MR_LVESV_AGES_Pt1.R (MR analysis; Outcomes: LVESV; GWAS: AGES)
• SDOH_MR_LVEDV_Science_Pt1.R (MR analysis; Outcomes: LVEDV; GWAS: Science)
• SDOH_MR_LVEDV_Fenland_Pt1.R (MR analysis; Outcomes: LVEDV; GWAS: Fenland)
• SDOH_Proteomics_Database_MR.Rmd (MR analysis with candidate proteomic markers)
• SDOH_MR_HF_Science_Pt1.R (MR analysis; GWAS: Science)
• SDOH_MR_HF_Fenland_Pt1.R (MR analysis; GWAS: Fenland)
• SDOH_MR_HF_AGES_Pt1.R (MR analysis; GWAS: AGES)

Mendelian Randomization/Plots

• Plots_CMR_MR_SDOH.Rmd (MR plot code)

Code/software

Software: R Studio (Version 2023.06.1+524)

Packages loaded: 

library(TwoSampleMR)

library(tidyverse)

library(ieugwasr)

library(openxlsx)

library(readxl)

library(magrittr)

library(tidyverse)

library(labelled)

library(haven)

library(mice)

library(survival)

library(broom)

library(tidyr)

library(glue)

library(furrr)

library(cowplot)

library(TwoSampleMR)

library(tidyverse)

library(ieugwasr)

library(openxlsx)

library(readxl)

library(TwoSampleMR)

library(coloc)

library(openxlsx)

library(readxl)

library(tidyverse)

library(readxl)

library(haven)

library(tidyverse)

library(openxlsx)

library(stringr)

library(magrittr)

library(tidyverse)

library(labelled)

library(haven)

library(mice)

library(survival)

library(broom)

library(tidyr)

library(glue)

library(furrr)

library(cowplot)

library(psych)

library(compareGroups)

library(openxlsx)

library(janitor)

library(gtsummary)

library(tidyverse)

library(survival)

library(broom)

library(survminer)

library(missMethods)

library(mice)

library(furrr)

Access information

Other publicly accessible locations of the data:

• UK Biobank is an open-access resource available to bona fide researchers undertaking health-related research in the public interest, subject to application and approval through the UK Biobank Access Management System. Approved researchers are provided de-identified data in accordance with UK Biobank access procedures. This study was performed under UK Biobank Application Number #285361.

Data was derived from the following sources:

• ARIC data are not publicly available due to participant privacy and consent limitations, but are available to qualified investigators through the NHLBI BioLINCC repository (https://biolincc.nhlbi.nih.gov/) or by application to the ARIC Coordinating Center, subject to institutional review and data use agreements. Derived variables and analytic datasets generated during the current study are available from the corresponding author with ARIC approval. 

Preprocessing required before running code

Before running the primary analysis, users should create an analysis-ready dataset by merging the demographic, SDOH, proteomics, clinical covariate, and heart failure outcome files by participant identifier. ("id") Participants with prevalent heart failure at baseline should be excluded. Analyses were conducted separately among Black and White participants, and the SDOH factor was derived within race strata using income, education, and Area Deprivation Index as detailed in the code deposited in Dryad.

Files should be placed in a local /data/ directory or the file paths in Source Functions/load-data.R should be modified accordingly. The data files format are wide, with one row per participant. 

The recommended script order is:

1. Source Functions/load-data.R
2. Primary Analysis/SDOH_Factor_Primary_Statistical_Analysis.Rmd
3. Mediation analysis scripts using Source Functions/MediationWithInteraction.R
4. UK Biobank replication script: UKB Replication/UKB Replication.Rmd
5. Mendelian randomization scripts in Mendelian Randomization/Running MR/
6. Colocalization script: Mendelian Randomization/SDOH Colocalization.Rmd
7. Plotting/table scripts in Source Functions/ and Mendelian Randomization/Plots/

All details regarding the required input variables, expected file structure, analytic dataset formatting, preprocessing steps, and recommended order for running the analysis scripts are provided in the code repository deposited in Dryad.

Human subjects data

ARIC data are not publicly available due to participant privacy and consent limitations, but are available to qualified investigators through the NHLBI BioLINCC repository (https://biolincc.nhlbi.nih.gov/) or by application to the ARIC Coordinating Center, subject to institutional review and data use agreements. Derived variables and analytic datasets generated during the current study are available from the corresponding author with ARIC approval. ARIC data made available to investigators are de-identified in accordance with participant privacy protections. Use of these data remains subject to institutional review and applicable data use agreements.