**OSF_EatingDisorders_COVID.Rdata** contains the data for incidence and outcomes of eating disorders during the COVID-19 pandemic. Loading the Rdata object in R will load a workspace with 25 variables: 16 containing incidence data, 8 containing data for the robustness analysis, and one containing the outcome data.
**Incidence data**
16 of these variables correspond to the incidence over each of the 4 years (2017, 2018, 2019, and 2020 with the latter corresponding to the pandemic) of the 4 categories of eating disorders: any eating disorder (ED), anorexia nervosa (AN), bulimia nervosa (BN), and eating disorder not otherwise specified (EDNOS).
Their name have the form `resYY_D` where `YY` is the year (17 to 20) and `D` is the disorder. For instance, `res20_AN` corresponds to the incidence of anorexia nervosa in 2020 (i.e. during the pandemic).
Each of these variables is a list containing the following elements (note: the use of the name ipr for variables is unfortunate as these represent incidence rates, not incidence prevalence ratio which is a different concept in epidemiology):
- `ipr`: the incidence by age group, sex, and time of the year
- `ipr.upper`: the upper bound of the 95% CI of the incidence
- `ipr.lower`: the lower bound of the 95% CI of the incidence
- `ipr.all`: the incidence across all ages and sexes
- `ipr.all.upper`: the upper bound of the 95% CI for ipr.all
- `ipr.all.lower`: the lower bound of the 95% CI for ipr.all
- `pos`: the number of cases by age group, sex, and time of the year
- `neg`: the number of non-case by age group, sex, and time of year
See also the accompanying R script for a user-friendly way of using this data.
**Robustness analysis data**
The variables with names `robustnessYY_D` are equivalent to the incidence data but pertain to the robustness analysis in which only patients who made a visit to a healthcare organisation after the study time window were used (see paper for details).
**Outcome data**
`resOutcomes` contains the results for the analysis of outcomes (in terms of suicidal ideation, suicidal attempt, and death) of patients diagnosed during the pandemic vs. before.
`resOutcomes` contains the following elements:
- `cohort1` is set to 'Diagnosis during the pandemic' which implies that `values1` (see below) refers to the outcomes for patients diagnosed during the pandemic.
- `cohort2` is set to 'Diagnosis before the pandemic'
- `outcomeNames`: the name of the outcomes
- `outcomes`: an array with the results for each outcome
The results of the Kaplan-Meier analysis for the *i*th outcome are stored in `outcomes[[i]]$KM` which contains the following elements:
- `endSurvivals`: the survival (1-event probability) probability at the end of the follow-up period for the two cohorts
- `Chi2`: the Chi squared statistics for the statistical test
- `p`: the corresponding p-value
- `survival$time`: a vector of time in days corresponding to the x-axis of all KM figures
- `survival$values1`: the survival (1-event probability) for the first cohort
- `survival$values2`: the survival (1-event probability) for the second cohort
- `survival$CI1`: the 95% confidence interval for the survival for the first cohort
- `survival$CI2`: the 95% confidence interval for the survival for the second cohort
- `HR`: the hazard ratio for the comparison
- `HR_ci`: the 95% CI for the hazard ratio
As an illustration the following few lines of code would plot the outcome probability for "Suicidal ideation" (`outcomes[[3]]`) for patients diagnosed during the pandemic (i.e. first cohort of the comparison so using `values1`):
`load('OSF_EatingDisorders_COVID.Rdata')`
`kmResults=resOutcomes$outcomes[[3]]$KM$survival`
`plot(kmResults$time,1-kmResults$values1,type='l')`
