Simulating data and analysing associations between school performance and psychiatric diagnoses

Introduction

This document shows how to simulate data, perform the analyses and vizualize the results as described in the article ‘School Performance and Later Diagnoses of Non-Affective Psychoses, Bipolar Disorder, and Depression’ published in Acta Psychiatrica Scandinavica (DOI:10.1111/acps.13481): http://doi.org/10.1111/acps.13481.

Load functions

Clone the repository from https://github.com/davgyl/school_sim.git and load the functions from the R-scripts.

Simulate data

source("R/dm01_utils.R")
source("R/dm02_pkg.r")
source("R/dm03_sim_data.R")
# Show simulated data
df_tot

## # A tibble: 60,000 × 24
##        pid d_start    Average covar_1 covar_2 covar_3 covar_4 covar_5 covar_6
##      <dbl> <date>       <dbl>   <dbl>   <dbl>   <dbl>   <dbl>   <dbl>   <dbl>
##  1 1000001 2003-05-31    8          0       0       0       0       1       0
##  2 1000002 2003-05-31    9.67       0       0       0       1       0       0
##  3 1000003 2003-05-31    5.17       0       1       0       1       0       1
##  4 1000004 2003-05-31    8.83       0       0       0       1       0       1
##  5 1000005 2003-05-31    7.67       0       0       0       0       0       0
##  6 1000006 2003-05-31    6.33       0       1       0       1       1       1
##  7 1000007 2003-05-31    5.33       0       0       0       0       0       1
##  8 1000008 2003-05-31    6.83       0       0       1       1       1       1
##  9 1000009 2003-05-31    8.5        0       0       0       0       0       1
## 10 1000010 2003-05-31    8.17       0       0       0       0       1       0
## # … with 59,990 more rows, and 15 more variables: subject_1 <dbl>,
## #   subject_2 <dbl>, subject_3 <dbl>, subject_4 <dbl>, subject_5 <dbl>,
## #   subject_6 <dbl>, event_dg1 <dbl>, event_dg2 <dbl>, event_dg3 <dbl>,
## #   fas_dg <chr>, time_dg1 <date>, time_dg2 <date>, time_dg3 <date>,
## #   dob <date>, fas <dbl>

Descriptive analyses of simulated data

source("R/mv01_descr_time_event.r")
table_1_data %>% select(-km_fit)

## # A tibble: 258 × 10
## # Groups:   outcome, chracteristic [63]
##    outcome chracteristic value     n n_risk  pyears events ci_estimate
##    <chr>   <chr>         <chr> <int>  <int>   <dbl>  <dbl>       <dbl>
##  1 dg1     all           all   60000  56733 676415.    796        4.90
##  2 dg2     all           all   60000  56407 674391.    470        1.05
##  3 dg3     all           all   60000  58734 688971.   2797        6.51
##  4 dg1     covar_1       0     58173  55041 656254.    769        5.16
##  5 dg1     covar_1       1      1827   1692  20161.     27        1.87
##  6 dg2     covar_1       0     58173  54714 654208.    442        1.02
##  7 dg2     covar_1       1      1827   1693  20183.     28        2.08
##  8 dg3     covar_1       0     58173  56962 668351.   2690        6.52
##  9 dg3     covar_1       1      1827   1772  20620.    107        6.24
## 10 dg1     covar_2       0     53145  50249 599795.    580        5.63
## # … with 248 more rows, and 2 more variables: ci_conf.high <dbl>,
## #   ci_conf.low <dbl>

summary_age

## # A tibble: 5 × 4
##   time_dg1 time_dg2 time_dg3 measure
##      <dbl>    <dbl>    <dbl> <chr>  
## 1    22.2     22.1     22.2  mean   
## 2     3.56     3.70     3.59 sd     
## 3    15.5     15.5     15.6  min    
## 4    22.3     22.2     22.1  median 
## 5    28.6     28.7     28.8  max

Modelling and visualization

To model the associations, run R/mv03_fit_models.r. To visualize the associations, run R/mv04_viz.r. To further edit tables and figres, run R/mv05_tabs_xtra_viz.r.

Versions

R version: 3.6.3. Version of packages are shown in the table below.

Package	Version
tidyverse	1.3.0
openxlsx	4.1.4
modelr	0.1.6
cowplot	1.0.0
GGally	1.5.0
survival	3.1.8
mgcv	1.8.31
lubridate	1.7.4
broom	0.5.5
sf	0.8.1
knitr	1.28
gridExtra	2.3
simstudy	0.1.15