============================================================================================================
START readme
============================================================================================================

This file focuses more on the details of the data package. For 'administrative' metadata on this data,
please view the metadata file within the parent folder of this data package (where this readme file
is also found).


============================================================================================================
## General
============================================================================================================

+ Author(s): L. Heerkens
+ Project: Fatty Liver Index and mortality after myocardial infarction
+ Contact: datasteward.hnh@wur.nl


============================================================================================================
## Title
============================================================================================================


"Data underlying the publication "Fatty Liver Index and mortality after myocardial infarction: A prospective analysis in the Alpha Omega Cohort" 


============================================================================================================
## Methods
============================================================================================================

# Introduction

Accumulating evidence shows that NAFLD might play a role in the etiology and progression of CVD, but little is known on the association of NAFLD and CVD mortality in patients with a history of a myocardial infarction (MI). Therefore, we studied the relationship of Fatty Liver Index (FLI), as indicator for non-alcoholic fatty liver disease (NAFLD), with 12-year risk of cardiovascular disease (CVD) and all-cause mortality in post-MI patients. We included 4165 Dutch patients from the Alpha Omega Cohort aged 60-80 years who had an MI ≤10 years prior to study enrolment. NAFLD was defined as FLI ≥60. Patients were followed for cause-specific mortality from enrolment (2002-2006) through December 2018. Hazard ratios for CVD and all-cause mortality were obtained by multivariable Cox regression using FLI <30 (indicating absence of NAFLD) as the reference. Baseline FLI as a continuous measure were studied with mortality using restricted cubic splines analyses. The median (IQR) FLI was 68 (48-84). Sixty percent of the patients had FLI ≥60, who were more likely to be male and more often had diabetes, high blood pressure and high serum cholesterol levels. During 12 years of follow-up, 2042 deaths occurred of which 846 from CVD. Patients with NAFLD were at increased risk of CVD mortality (HR: 1.55 [1.19, 2.03]) and all-cause mortality (HR: 1.21 [1.03; 1.41]) compared to patients without NAFLD. Results remained consistent after excluding patients with obesity and diabetes. To conclude, the adverse association of FLI with CVD mortality were stronger in female than in male patients with conventional cut-off points. FLI ≥60, indicating NAFLD, was a predictor for CVD and all-cause mortality in post-MI patients, independent of other cardiometabolic risk factors. However, cut-off points might differ between male and female patients for predicting CVD mortality.


# Measurements and data collection

At baseline (2002-2006), the Alpha Omega Cohort comprised 4837 patients who experienced a myocardial infarction ≤10 years prior to their study enrolment. Patients missing data on FLI components (n=378) were excluded. Patients with missing data on alcohol consumption (n=200) or excessive alcohol consumption (>30 grams/day for men or >20 grams/day for women, n=344) were excluded as well, resulting in a population of 4165 patients for studying baseline FLI and mortality risk. Measurements were repeated in a random subset of the cohort which was re-examined after approximately 40 months (n=2503) After applying the same exclusion criteria, 1678 patients were available for the analysis of 40-month change in FLI and mortality risk.

FLI was calculated at baseline and after 40 months of follow-up on the basis of BMI (kg/m2), waist circumference (cm), fasted and non-fasted serum GGT (U/L), and fasted or non-fasted serum triglycerides (mg/dL). Physical examination was performed in the hospital or at the patients’ homes by trained research nurses. Weight and height were measured, and BMI was calculated as body weight divided by square height (kg/m2). Waist circumference (cm) was measured at the midpoint between the bottom rib and the top of the hipbone using a non-elastic tape. Venous blood samples (30 mL) were drawn fasted (>8 hours, 35% of the population for analysis) or non-fasted and were sent to the laboratory by next-working-day mail service. Upon arrival, blood samples were immediately processed and stored at -80 degrees Celsius. Serum triglycerides were determined by standard assays (Roche Diagnostics; cat. no. 1488872) on an automated analyzer (Hitachi 912; Roche Diagnostics) with an inter- and intra-assay coefficient of variation <10%. Fasted and non-fasted serum GGT was determined by standard assays (Abbott Diagnostics; cat. no. 7D6522) on an automated analyzer (ARCHITECT ci8200; Abbott) with an intra- and inter-assay coefficient of variation <10%. 
We used the conventional cut-off points of FLI <30 to indicate the absence of hepatic steatosis with a probability of 87% and FLI ≥60 to indicate the presence of hepatic steatosis. These cut-off points have been validated in a Caucasian population-based cohort showing a probability of 87% for absence 86% for presence of NAFLD,  and is calculated as follows:
FLI = (e 0.953*loge (triglycerides) + 0.139*BMI + 0.718*loge (GGT) + 0.053*waist circumference - 15.745) / (1 + e 0.953*loge (triglycerides) + 0.139*BMI + 0.718*loge (GGT) + 0.053*waist circumference - 15.745) * 100

The primary endpoints of the current study were CVD mortality and all-cause mortality. Patients were monitored for their vital status from baseline through 31 December 2018 through linkage with municipal registries. Data collection on cause-specific mortality occurred in three phases. During the Alpha Omega Trial (2002-2009), information was obtained from the national mortality registries [Statistics Netherlands (CBS)], treating physicians, and close family members. Primary and contributing causes of death were coded by an independent Endpoint Adjudication Committee and described in detail elsewhere.[12, 13] After the trial through 2012, mortality data were obtained from CBS for primary and contributing causes of death. From 2013 onwards, data on only primary cause of death were obtained from CBS. Treating physicians filled out an additional cause-of-death questionnaire (response rate: 67%), which was coded by study physicians who were not involved in the current analysis. The endpoint CVD was allocated to all patients for whom it was a primary or contributing cause of death, based on any of the data sources.
Fatal endpoints were coded according to the International Classification of Diseases, 10th revision (ICD-10). CVD mortality comprised coronary heart diseases (codes I20-I25), cardiac arrest (I46), heart failure (I50), stroke (I60-I69), and undefined sudden death (R96). 

Data at baseline and after 40 months of follow-up were collected on demographics, smoking status, alcohol consumption and medication use. 
Self-reported smoking status was assessed in four categories (current, quit ≤10 years, quit >10 years, never). Self-reported medication use was checked by research nurses and coded according to the Anatomical Therapeutic Chemical Classification System (ATC). The ATC code for statins was C10A, A10 for antidiabetic medication, and C02, C03, C07, C08, and C09 for antihypertensives. The intake of alcoholic beverages (frequency, amount) over the past month was assessed using a Food Frequency Questionaire, from which total ethanol intake (g/d) was computed. Alcohol consumption in this population with non-excessive alcohol intake (< 20 g/d in female or < 30 g/d in male) was categorized as follows: abstinence (0 g/d), light (>0-10 g/d), and moderate (>10 g/d). Trained research nurses measured blood pressure at baseline and after 40 months of follow-up. Blood pressure was measured twice on each occasion after a 15-minute rest, and values were averaged. 
Serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) were analyzed in stored blood by standard assays (Abbott Diagnostics; cat. no. 8L9222 and 8L9122) on an automated analyzer (ARCHITECT ci8200; Abbott) inter-assay coefficient of variation <10%, respectively. Prevalent diabetes was assessed by physician’s diagnosis, use of diabetic medication or when patient had elevated plasma glucose (≥7.0 mmol/L if fasted >4 h or ≥11.0 mmol/L if not fasted 

Baseline characteristics are presented across categories of FLI (<30; ≥30-<60, ≥60) as means ± SD for normally distributed data, medians and interquartile range (IQR) for skewed data, and n (%) for categorical data. 
Cox proportional hazards models were used to estimate hazard ratios (HRs) with 95% confidence intervals (95% CIs) for CVD mortality, CHD mortality, and all-cause mortality. In the main models, FLI at baseline was analyzed in categories, using FLI <30 as the reference category. The assumption of proportionality was checked visually using a log-minus-log plot of survival and time and was met. Survival time (in person-years) was calculated from the date of enrolment to date of death or end of follow-up. One person was lost to follow-up and censored after 2.9 years.
Unadjusted HRs were presented in model 1. Subsequently, HRs were adjusted for age and sex (model 2). Model 3 also included systolic blood pressure (mmHg), smoking status (4 categories), alcohol consumption (3 categories), statin use (yes/no), time since last MI (years), fasting, and the type of intervention during the Alpha Omega Trial (types of omega-3 fatty acids or placebo; 4 categories). The latter, however, was not a confounder because it was randomly assigned. Missing data for smoking status (n=1), systolic blood pressure (n=6), and time since last MI (n=41) were imputed with a sex-specific mode, mean or median, respectively.
Restricted cubic splines (RCS) analyses were performed to assess the continuous association of FLI with CVD and all-cause mortality, using the fully adjusted model. FLI = 30 was set as the reference and four knots were placed at 5th, 35th, 65th, and 95th percentiles. The number of these knots were chosen according to the lowest Akaike’s Information Criteria. The Wald chi-square test was performed to test for nonlinearity. RCS analyses were repeated and stratified by sex to assess effect measure modification. 
We performed a series of additional analyses. We stratified the main analysis, studying FLI categories, for sex and age. Furthermore, we repeated the main analysis in non-obese (n=3163) and non-diabetic (n=3308) patients to study FLI ≥60 as a predictor of CVD and all-cause mortality independent of obesity and diabetes. Analyses were performed for single FLI components (expressed as Z-scores) and AST/ALT ratio, as an additional diagnostic marker for liver diseases, in relation to CVD and all-cause mortality. Furthermore, we repeated the main analysis using FLI <60 as the reference category. Sensitivity analyses were performed by excluding the first two years of follow-up and repeating the main analysis using the first 10 years of follow-up. 
In our cohort of patients who underwent an assessment 40 months after baseline, we repeated RCS analyses to study the change of FLI after 40 months of follow-up (baseline FLI – FLI after 40 months of follow-up; expressed as Z-scores) in relation to CVD and all-cause mortality, excluding the first 40 months of follow-up. Zero change of FLI was set as the reference and the analyses were additionally stratified for NAFLD at baseline (FLI <60; and FLI ≥60).
Two-sided P-values <0.05 were considered statistically significant. 
============================================================================================================
## FolderStructure
============================================================================================================

There is only one parent directory present containing all files.


============================================================================================================
## FolderContents
============================================================================================================

[If you're presenting only individual files without any folder structure or sub-folders, please only
use the first entry ('- parent_folder/'). Example (delete this explanation and example):

- parent_folder/
[minimal dataset to reproduce results with a corresponding codebook]




============================================================================================================
## Software
============================================================================================================

SoftwareRequired: 

SAS 9.4 and R version 4.1.0


OtherSoftwareRequirements: 

used R packages:
-Tableone
-haven
-dplyr
-grid
-forestplot
-rms
-survival
-gmodels

============================================================================================================
## FileFormats
============================================================================================================


.sav
.sas7bdat


============================================================================================================
## CodeBook
============================================================================================================

Codebook, named 'Codebook_fli_mortality_mipatients.xlsx', can be found in the parent folder.


============================================================================================================
## Other
============================================================================================================

[describe any other attention points that will help understandability of your data package; delete this 
explanation]


============================================================================================================
END readme
============================================================================================================
