Homework 5: Potential Determinants of Depression in the United Kingdom
2025-06-04
New section: “Predictors of Clinically Significant Depression”
# Subset UK data (rows where country is "United Kingdom", all columns)
df_uk = df[df$cntry == "United Kingdom", ]# Convert columns in df_uk (these are "factors")
# Convert them to real numeric values using as.numeric, as.character
df_uk$d20 = as.numeric(as.character(df_uk$fltdpr))
df_uk$d21 = as.numeric(as.character(df_uk$flteeff))
df_uk$d22 = as.numeric(as.character(df_uk$slprl))
df_uk$d23 = as.numeric(as.character(df_uk$wrhpp))
df_uk$d24 = as.numeric(as.character(df_uk$fltlnl))
df_uk$d25 = as.numeric(as.character(df_uk$enjlf))
df_uk$d26 = as.numeric(as.character(df_uk$fltsd))
df_uk$d27 = as.numeric(as.character(df_uk$cldgng))
# Reverse code d23 and d25 (scoring 0-4; 5s NA)
df_uk$d23 = 5 - df_uk$d23
df_uk$d25 = 5 - df_uk$d251 Introduction
Depression is a prevalent global health issue, experienced by 4-10% of the global population over their lifetime (Chapman et al., 2022). With currently, around 280 million people (3.8%) being affected globally (WHO, 2023), depression ranks among the top contributors to the global health burden (Santomauro et al., 2021). Although adequate and cost-efficient treatments are available across low-, middle-, and high-income countries, and stigma surrounding professional help is decreasing, depression treatment challenges are common occurrences (Ferrari et al., 2024); delayed diagnosis, treatments that are not tailored to individual needs, and inadequate responses to treatment attempts represent such obstacles (Chapman et al., 2022; Ferrari et al., 2024). For instance, in the United Kingdom (UK), 15 30% of individuals still suffer from depression after undergoing two or more treatments. This is relevant, as depression affects various aspects of human’s life, including education, employment, and personal relationships (OECD & European Commission, 2024), while contributing significantly to economic and healthcare costs (Statista, 2024; Vinokur et al., 1996).
1.1 Hypotheses concerning the UK:
H1. The prevalence of depression increases with experienced discrimination based on an individual’s sexuality.
H2. The prevalence of depression increases with experienced discrimination based on an individual’s skin colour or race.
H3. The prevalence of depression decreases with age.
H4. The prevalence of depression among females compared to males is higher.
2 Methods
The present paper aimed to investigate depression in a British population, as 15-30% of individuals do not recover from depression after two or more treatments (Chapman et al., 2022) and therefore a greater understanding of potential contributing factors is crucial for improving recovery outcomes.
2.1 Dependent Variable
Our dependent Variable is “Depression”, which was measured using the Centre for Epidemiological Studies Depression Scale (CES-D8), an 8-item scale.
2.2 Independent Variables
Discrimination based on the respondent’s sexuality (nominal: “marked”, “not marked”), discrimination based on colour or race (nominal: “marked”, “not marked”), age (ratio scale: 15-90), and gender (nominal: “male”, “female”) were considered as potential factors influencing depression.
2.3 Data Analysis
To assess the internal consistency of the depression scale, Cronbach’s alpha was calculated, typically ranging from 0 to 1 (Döring & Bortz, 2016), although it can sometimes be negative (Bühner, 2005).
# calculation of Cronbach's alpha (using df_uk) to check internal consistency ("reliability") of depression items
cronbach.alpha(df_uk[,c("d20", "d21", "d22", "d23", "d24", "d25", "d26", "d27")], na.rm=T)##
## Cronbach's alpha for the 'df_uk[, c("d20", "d21", "d22", "d23", "d24", "d25", "d26", "d27")]' data-set
##
## Items: 8
## Sample units: 1684
## alpha: NAThe degree of agreement between items was high, with an alpha of NA, well below the recommended threshold of 0.7 (Hair, 2010), indicating that not all items measure the same underlying construct of depression (Osburn, 2000).
3 Results
3.1 Sample Description
# rows where country is "United Kongdom", all columns
df_uk = df[df$cntry == "United Kingdom", ]# Convert relevant columns to numeric in df_uk
df_uk$d20 = as.numeric(df_uk$fltdpr)
df_uk$d21 = as.numeric(df_uk$flteeff)
df_uk$d22 = as.numeric(df_uk$slprl)
df_uk$d23 = as.numeric(df_uk$wrhpp)
df_uk$d24 = as.numeric(df_uk$fltlnl)
df_uk$d25 = as.numeric(df_uk$enjlf)
df_uk$d26 = as.numeric(df_uk$fltsd)
df_uk$d27 = as.numeric(df_uk$cldgng)
# Calculate depression score
df_uk$depression = rowSums(df_uk[, c("d20", "d21", "d22", "d23", "d24", "d25", "d26", "d27")], na.rm = FALSE) / 8
# Check descriptive statistics
summary(df_uk$depression)## Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
## 1.000 1.750 1.875 1.952 2.125 4.000 49The dataset was subset to include only participants from the United Kingdom, resulting in an initial sample size of n=1684. However, the final sample consisted of n= 1635 valid respondents, aged 15 to 90 years, for the depression variable in the UK sample.
3.2 Calculation of the Depression Score
# calculation of the average score
df_uk$depression = rowSums(df_uk[, c("d20", "d21", "d22", "d23", "d24", "d25", "d26", "d27")]) / 8This section presents information about the depression score.
3.2.1 Descriptive Statistics of Depression
summary(df_uk$depression)## Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
## 1.000 1.750 1.875 1.952 2.125 4.000 49# minimum value of 1, a maximum of 4, a median of 1.875, and a mean of 1.952 (right-skewed distribution)# data.frame to make a table out of it
summary_df = data.frame(
Statistic = c("Minimum", "Maximum", "Median", "Mean"),
Value = c(1, 4, 1.875, 1.952)
)# table
kable_styling(
kable(summary_df,
col.names = c("Statistic", "Value"),
caption = "Summary Statistics of Depression Scores")
)| Statistic | Value |
|---|---|
| Minimum | 1.000 |
| Maximum | 4.000 |
| Median | 1.875 |
| Mean | 1.952 |
The depression scale has a range from 1 (lowest depression level) to 4 (highest depression level). Based on the data, we observe the following: At least one individual answered all items with a score of 1 (indicating the lowest possible depression level), and at least one individual answered all items with a score of 4 (indicating the highest possible depression level). The majority of participants report low to moderate depression levels, as indicated by the median (1.875) and mean (1.952) scores, which are relatively low. This suggests that most participants fall within the lower half of the depression scale.
3.2.2 Frequency Distribution of Depression
Here is a comprehensive overview of the frequency distribution:
table(df_uk$depression)##
## 1 1.125 1.25 1.375 1.5 1.625 1.75 1.875 2 2.125 2.25 2.375 2.5
## 11 1 6 17 110 178 297 304 221 176 97 72 48
## 2.625 2.75 2.875 3 3.125 3.25 3.375 3.75 4
## 25 29 13 10 6 9 3 1 1# frequency distribution of the new variable (depression)# now we need metric information
# USE DUMMY CODING for "dscrsex", "dscrrce", "gndr"
## discrimination, sexuality (dscrsex): 0 = not marked, 1 = marked
## discrimination, skin colour or race (dscrrce): 0 = not marked, 1 = marked
## gender (gndr): 0 = male, 1 = female
# create dummy variable: marked (discrimination based on sexuality was perceived or reported)
# frequency distribution of "dscrsex"
table(df_uk$dscrsex)##
## Not marked Marked
## 1646 38df_uk$marked = NA #initialize variable with NA
df_uk$marked[df_uk$dscrsex=="Not marked"] = 0
df_uk$marked[df_uk$dscrsex=="Marked"] = 1
# create dummy variable: marked (discrimination based on skin colour or race was perceived or reported)
# frequency distribution of "dscrrce"
table(df_uk$dscrrce)##
## Not marked Marked
## 1623 61df_uk$marked = NA #initialize variable with NA
df_uk$marked[df_uk$dscrrce=="Not marked"] = 0
df_uk$marked[df_uk$dscrrce=="Marked"] = 1
# create dummy variable: female
# frequency distribution of "gndr"
table(df_uk$gndr)##
## Male Female
## 824 860df_uk$female = NA #initialize variable with NA
df_uk$female[df_uk$gndr=="Male"] = 0
df_uk$female[df_uk$gndr=="Female"] = 13.3 Create Likert Table and Likert Plot
# variables: How often during the past week...
# fltdpr: felt depressed
# flteeff: felt that everything they did was an effort
# slprl: sleep was restless
# wrhpp: were happy
# fltlnl: felt lonely
# enjlf: enjoyed life
# fltsd: felt sad
# cldgng: could not get going (felt lethargic and lacked motivation)
vnames = c("fltdpr", "flteeff", "slprl", "wrhpp", "fltlnl", "enjlf", "fltsd", "cldgng")
vnames## [1] "fltdpr" "flteeff" "slprl" "wrhpp" "fltlnl" "enjlf" "fltsd"
## [8] "cldgng"likert_df = df[,vnames] # recreate a new dataframe, d.h., df[rows comma colum].3.3.1 Create Basic Frequencies
# create table directly from long format data
likert(likert_df)## Item None or almost none of the time Some of the time Most of the time
## 1 fltdpr 64.915835 29.06631 4.557165
## 2 flteeff 48.395568 38.42383 9.814171
## 3 slprl 43.873854 39.87056 11.625059
## 4 wrhpp 4.003510 23.53973 48.886939
## 5 fltlnl 68.136458 24.27532 5.302253
## 6 enjlf 5.338783 24.82572 44.804153
## 7 fltsd 52.489933 41.07451 4.859808
## 8 cldgng 55.673484 36.10353 6.217928
## All or almost all of the time
## 1 1.460694
## 2 3.366431
## 3 4.630532
## 4 23.569817
## 5 2.285972
## 6 25.031346
## 7 1.575748
## 8 2.0050563.3.2 Create a Basic Plot
# create plot directly from long format data
plot(likert(likert_df))
# Append mean and counts
# Append mean
# -> convert to numeric
# long version
likert_numeric_df = df[,vnames]
likert_numeric_df$d20 = as.numeric(likert_numeric_df[,vnames[1]])
likert_numeric_df$d21 = as.numeric(likert_numeric_df[,vnames[2]])
likert_numeric_df$d22 = as.numeric(likert_numeric_df[,vnames[3]])
likert_numeric_df$d23 = as.numeric(likert_numeric_df[,vnames[4]])
likert_numeric_df$d24 = as.numeric(likert_numeric_df[,vnames[5]])
likert_numeric_df$d25 = as.numeric(likert_numeric_df[,vnames[6]])
likert_numeric_df$d26 = as.numeric(likert_numeric_df[,vnames[7]])
likert_numeric_df$d27 = as.numeric(likert_numeric_df[,vnames[8]])
# get means
# long version
likert_means = c()
likert_means$d20 = mean(likert_numeric_df$d20, na.rm=T)
likert_means$d21 = mean(likert_numeric_df$d21, na.rm=T)
likert_means$d22 = mean(likert_numeric_df$d22, na.rm=T)
likert_means$d23 = mean(likert_numeric_df$d23, na.rm=T)
likert_means$d24 = mean(likert_numeric_df$d24, na.rm=T)
likert_means$d25 = mean(likert_numeric_df$d25, na.rm=T)
likert_means$d26 = mean(likert_numeric_df$d26, na.rm=T)
likert_means$d27 = mean(likert_numeric_df$d27, na.rm=T)# Append counts
# get counts
# long version
likert_counts = c()
likert_counts$d20 = sum(!is.na(likert_numeric_df$d20))
likert_counts$d21 = sum(!is.na(likert_numeric_df$d21))
likert_counts$d22 = sum(!is.na(likert_numeric_df$d22))
likert_counts$d23 = sum(!is.na(likert_numeric_df$d23))
likert_counts$d24 = sum(!is.na(likert_numeric_df$d24))
likert_counts$d25 = sum(!is.na(likert_numeric_df$d25))
likert_counts$d26 = sum(!is.na(likert_numeric_df$d26))
likert_counts$d27 = sum(!is.na(likert_numeric_df$d27))# append means and counts to table
likert_table = likert(likert_df)$results # we save the "inner" data frame of the likert structure ...
likert_table$Mean = unlist(likert_means) # ... and append new columns to the data frame
likert_table$Count = unlist(likert_counts)# set new item labels
likert_table$Item = c(
d20="How often during the last week participants felt depressed",
d21="How often during the last week participants felt that everything they did was an effort",
d22="How often during the last week participants's sleep was restless",
d23="How often during the last week participants were happy",
d24="How often during the last week participants felt lonely",
d25="How often during the last week participants enjoyed life",
d26="How often during the last week participants felt sad",
d27="How often during the last week participants could not get going (felt lethargic and lacked motivation)"
)# round all percetage values to 1 decimal digit
likert_table[,2:5] = round(likert_table[,2:5],1)
# round means to 3 decimal digits
likert_table[,6] = round(likert_table[,6],3)# create formatted table
kable_styling(kable(likert_table,
caption = "Distribution of Answers Regarding Depression (ESS Round 11, Based on Data From the United Kingdom)"
)
)| Item | None or almost none of the time | Some of the time | Most of the time | All or almost all of the time | Mean | Count |
|---|---|---|---|---|---|---|
| How often during the last week participants felt depressed | 64.9 | 29.1 | 4.6 | 1.5 | 1.426 | 39981 |
| How often during the last week participants felt that everything they did was an effort | 48.4 | 38.4 | 9.8 | 3.4 | 1.682 | 39983 |
| How often during the last week participants’s sleep was restless | 43.9 | 39.9 | 11.6 | 4.6 | 1.770 | 40017 |
| How often during the last week participants were happy | 4.0 | 23.5 | 48.9 | 23.6 | 2.920 | 39890 |
| How often during the last week participants felt lonely | 68.1 | 24.3 | 5.3 | 2.3 | 1.417 | 39983 |
| How often during the last week participants enjoyed life | 5.3 | 24.8 | 44.8 | 25.0 | 2.895 | 39878 |
| How often during the last week participants felt sad | 52.5 | 41.1 | 4.9 | 1.6 | 1.555 | 39981 |
| How often during the last week participants could not get going (felt lethargic and lacked motivation) | 55.7 | 36.1 | 6.2 | 2.0 | 1.546 | 39949 |
# create basic plot (code also valid)
plot(likert(summary=likert_table[,1:4])) # limit to columns 1:4 to skip mean and count
3.4 Multivariate Regression (based on UK data) - Unweighted
# final model and show extended summary (unweighted)
# 824 male (48,93%); 860 female (51,67%)
# = 0.5 / (824/1684) ≈ 0.5 / 0.4893 ≈ 1.022 (male)
# = 0.5 / (860/1684) ≈ 0.5 / 0.5107 ≈ 0.979 (female)
model = lm(depression ~ dscrsex + female, data=df_uk)
summary(model)##
## Call:
## lm(formula = depression ~ dscrsex + female, data = df_uk)
##
## Residuals:
## Min 1Q Median 3Q Max
## -0.98402 -0.23402 -0.03865 0.14098 2.01598
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.91365 0.01259 152.029 < 2e-16 ***
## dscrsexMarked 0.08674 0.05793 1.497 0.135
## female 0.07038 0.01746 4.030 5.83e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.3526 on 1632 degrees of freedom
## (49 observations deleted due to missingness)
## Multiple R-squared: 0.0109, Adjusted R-squared: 0.009691
## F-statistic: 8.995 on 2 and 1632 DF, p-value: 0.00013033.5 Multivariate Regression (based on UK data) - Weighted
# final model and show extended summary (weighted)
df_uk$weight = ifelse(df_uk$female == 1, 0.979, 1.022)
model = lm(depression ~ dscrsex + female, data = df_uk, weights = weight)
summary(model)##
## Call:
## lm(formula = depression ~ dscrsex + female, data = df_uk, weights = weight)
##
## Weighted Residuals:
## Min 1Q Median 3Q Max
## -0.97367 -0.23159 -0.03913 0.13946 1.99466
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.91370 0.01245 153.761 < 2e-16 ***
## dscrsexMarked 0.08485 0.05773 1.470 0.142
## female 0.07035 0.01745 4.032 5.78e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.3524 on 1632 degrees of freedom
## (49 observations deleted due to missingness)
## Multiple R-squared: 0.01087, Adjusted R-squared: 0.009658
## F-statistic: 8.967 on 2 and 1632 DF, p-value: 0.0001339A linear regression was conducted to predict depression scores based on dscrsexMarked (reported or experienced disrimination because of one’s sex) and female variables. Intercept: is estimated at 1.91365 (p < 2e-16, highly significant).
dscrsexMarked: This variable has a coefficient estimate of 0.08674 but is not statistically significant (p = 0.135), suggesting no strong evidence that this variable predicts depression in this model.
female: The coefficient is 0.07038, statistically significant (p = 5.83e-05), indicating that females tend to have slightly higher depression scores by about 0.07 units compared to the reference group.
The overall model explains a small proportion of variance in depression scores: Multiple R-squared = 0.0109 (about 1.1% variance explained), which means, that there must be additional factors influencing depression that are not included in this model.
3.6 Predictors of Clinically Significant Depression
In this work, the cut-off score of 9 (out of 24, considering 8 items scored 0–3) has been considered to identify whether individuals show a clinical level of depression. This choice is based on findings by Briggs et al. (2018), stating that, firstly, a cut-off score of 9 (based on the sum of the scores from all 8 items on the CES-D-8) identifies clinically significant depressive symptoms precisely. Secondly, at a cut-off score of 9, both sensitivity (98%) and specificity (83%) were high. Although the specificity is higher, for example at a cut-off score of 11 (96%), the sensitivity with 95% is lower. Regarding the question of whether an individual shows clinical-level depression, it can be argued that to detect as many real depression cases as possible (high sensitivity) while also keeping false positives reasonably low (good specificity), a cut-off score of 9 is a meaningful choice. However, as in our data the scoring system is 0–4, and to preserve the same threshold for identifying clinical-level depression, a higher cut-off of 12 (proportionally adjusted) has been decided upon (although lack of empirical validation of the new cut-off score).
Information for myself: High sensitivity = few false negatives. 98% of people who actually have the disease were correctly identified
# Binary dependent variable (clincial-level-depression = yes/no)
# Remember:
# How often during the past week...
# fltdpr: felt depressed
# flteeff: felt that everything they did was an effort
# slprl: sleep was restless
# wrhpp: were happy
# fltlnl: felt lonely
# enjlf: enjoyed life
# fltsd: felt sad
# cldgng: could not get going (felt lethargic and lacked motivation)'
# Calculate CES-D8 total score (ranging 0 to 32) for UK data
# Remove rows with missing values in the 8 depression items
df_uk_complete = df_uk[complete.cases(df_uk[, c("d20","d21","d22","d23","d24","d25","d26","d27")]), ]
# Calculate total scores on complete cases only (1635 out of 1684)
df_uk_complete$totalscore = rowSums(df_uk_complete[, c("d20","d21","d22","d23","d24","d25","d26","d27")])
# Create binary variable: clinical-level depression
# Justification: As stated before, a cut-off score of 9 identifies clinical significant depressive symptoms accurately.
# Since here a 0–4 scale (instead of 0–3) is used, total score ranges from 0–32, not from 0-24.
# Proportional adjustment: 24 - 9; 32 - x --> 9/24*32 = 12
df_uk_complete$clinical_depression = ifelse(df_uk_complete$totalscore >= 12, 1, 0) # true (1); false (0)
# Report frequency and proportions
table(df_uk_complete$clinical_depression)##
## 0 1
## 35 1600# in percent
round(prop.table(table(df_uk_complete$clinical_depression)), 2)*100##
## 0 1
## 2 98# Interpretation:
# Out of 1635 valid answers from UK individuals (who experience depression), 1600 showed a clinical level of depression (98%), whereas 35 individuals did not show a clinical level of depression (2%).# Fit and interpret a logistic regression model using that binary outcome.
# Clinically depression symptoms YES or NO (dependent variable) (binary outcome)
# Transform the binary
# Create a model that allows us to predict with the explanatory variable "discrimination based on sexuality".
# Estimate regression model, ignore the warning
aModel = glm(clinical_depression ~ dscrsex, data=df_uk_complete, family=binomial)
# Show summary of regression model (we have the logarithm)
summary(aModel)##
## Call:
## glm(formula = clinical_depression ~ dscrsex, family = binomial,
## data = df_uk_complete)
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 3.8280 0.1733 22.083 <2e-16 ***
## dscrsexMarked -0.2171 1.0281 -0.211 0.833
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 338.33 on 1634 degrees of freedom
## Residual deviance: 338.29 on 1633 degrees of freedom
## AIC: 342.29
##
## Number of Fisher Scoring iterations: 6coef(aModel)## (Intercept) dscrsexMarked
## 3.8280018 -0.2170839# Interpretation:
# For a one-unit change in dscrsex (from no discrimination to discrimination based on sexuality),
# the log odds of clinical depression decreases by -0.2171. However, this effect is not statistically significant (p = 0.833).
# Now we need to dilogarithm it - derive the coefficient of this model
# Calculate the OR
exp(coef(aModel))## (Intercept) dscrsexMarked
## 45.9705881 0.8048624# Calculate Confidence Intervals for ORs
exp(confint(aModel)) ## 2.5 % 97.5 %
## (Intercept) 33.3028409 65.84588
## dscrsexMarked 0.1663652 14.49491# Interpretation:
# Odds ratio for discrsex (marked) (vs. not marked): OR for marked (yes) - experiencing discrimination based on sexuality compared to not marked (no) is 0.8, so those you agree have lower odds, which is contradictory. However, this effect is not statistically significant because the 97.5% CI ranges from 0.166 to 14.495, which includes 1.# Assess model fit via pseudo‐R-squared
r_mcfadden = with(summary(aModel), 1 - deviance/null.deviance)
r_nagelkerke = with(summary(aModel), r_mcfadden/(1 - (null.deviance / nrow(aModel$data)*log(2))))
r_mcfadden## [1] 0.0001236029r_nagelkerke## [1] 0.0001443002# Interpretation:
# The model with discrimination based on sexuality (dscrsex) as the only predictor explains very little of the variance in clinical depression.
# McFadden’s R-Squared = 0.00012 and Nagelkerke’s R-Squared = 0.00014, meaning less than 0.02% of the variance is explained.
# Therefore, dscrsex alone is not a meaningful predictor of clinical depression in this sample.
# Compared to previous linear regression analyses where discrimination had a significant effect (p < 0.05),
# this logistic regression model shows a weaker explanatory power, explaining approximately 0.93% of the variation,
# which is still very low. 3.6.1 Key Statements
Out of 1635 valid answers from UK individuals (who experience depression), 1600 showed a clinical level of depression (98%), whereas 35 individuals did not show a clinical level of depression (2%).
For a one-unit change in dscrsex (from no discrimination to discrimination based on sexuality), the log odds of clinical depression decreases by -0.2171. However, this effect is not statistically significant (p = 0.833).
Odds ratio for discrsex (marked) (vs. not marked): OR for marked (yes) - experiencing discrimination based on sexuality compared to not marked (no) is 0.8, so those you agree have lower odds, which is contradictory. However, this effect is not statistically significant because the 97.5% CI ranges from 0.166 to 14.495, which includes 1.
The model with discrimination based on sexuality (dscrsex) as the only predictor explains very little of the variance in clinical depression. McFadden’s R-Squared = 0.00012 and Nagelkerke’s R-Squared = 0.00014, meaning less than 0.02% of the variance is explained. Therefore, dscrsex alone is not a meaningful predictor of clinical depression in this sample.
Compared to previous linear regression analyses where discrimination had a significant effect (p < 0.05), explaining approximately 0.93% of the variation, this logistic regression model shows a weaker explanatory power. Very poor explanation.
4 References
Bühner, M. (2005). Einführung in die Test- und Fragebogenkonstruktion (Nachdr.). Pearson Studium.
Chapman, N., Browning, M., Baghurst, D., Hotopf, M., Willis, D., Haylock, S., Zakaria, S., Speechley, J., Withey, J., Brooks, E., Chan, F., Pappa, S., Geddes, J., Insole, L., Mohammed, Z., Kessler, D., Jones, P. B., Mansoori, P., & Difficult to Treat Depression Research Priority Setting Group. (2022). Setting national research priorities for difficult-to-treat depression in the UK between 2021-2026. Journal of Global Health, 12, 09004. https://doi.org/10.7189/jogh.12.09004
Döring, N., & Bortz, J. (with Pöschl-Günther, S.). (2016). Forschungsmethoden und Evaluation in den Sozial- und Humanwissenschaften (5. vollst. überarb. Aufl). Springer. 13
Ferrari, A. J., Santomauro, D. F., Aali, A., Abate, Y. H., Abbafati, C., Abbastabar, H., Abd ElHafeez, S., Abdelmasseh, M., Abd-Elsalam, S., Abdollahi, A., Abdullahi, A., Abegaz, K. H., Abeldaño Zuñiga, R. A., Aboagye, R. G., Abolhassani, H., Abreu, L. G., Abualruz, H., Abu-Gharbieh, E., Abu-Rmeileh, N. M., … Murray, C. J. L. (2024). Global incidence, prevalence, years lived with disability (YLDs), disability adjusted life-years (DALYs), and healthy life expectancy (HALE) for 371 diseases and injuries in 204 countries and territories and 811 subnational locations, 1990 2021: A systematic analysis for the Global Burden of Disease Study 2021. The Lancet, 403(10440), 2133–2161. https://doi.org/10.1016/S0140-6736(24)00757-8
OECD & European Commission. (2024). Health at a Glance: Europe 2024: State of Health in the EU Cycle. OECD. https://doi.org/10.1787/b3704e14-en
Santomauro, D. F., Mantilla Herrera, A. M., Shadid, J., Zheng, P., Ashbaugh, C., Pigott, D. M., Abbafati, C., Adolph, C., Amlag, J. O., Aravkin, A. Y., Bang-Jensen, B. L., Bertolacci, G. J., Bloom, S. S., Castellano, R., Castro, E., Chakrabarti, S.,
Vinokur, A. D., Price, R. H., & Caplan, R. D. (1996). Hard times and hurtful partners: How financial strain affects depression and relationship satisfaction of unemployed persons and their spouses. Journal of Personality and Social Psychology, 71(1), 166–179. https://doi.org/10.1037/0022-3514.71.1.166
WHO. (2023, March 31). Depressive disorder (depression). https://www.who.int/news room/fact-sheets/detail/depression