Final Paper: PNES Online Survey- Patient’s trust into medical workers, Subjective Well-being and the development of Depression
Lena Bareuther
- Februar 2016
Section I. Dataset Description
The present study and the associated dataset was conducted by myself: Patients with PNES (psychogenic non-epileptic seizures) were recruited via closed online support groups for psychogenic non-epileptic seizures from social networking service Facebook as a self-selected sample to examine patient’s trust into medical workers, their Subjective Well-being and the development of depression as a result of suffering from PNES for years. The present study was at two points in time promoted with a direct link to the online survey (https://www.soscisurvey.de/coiversiondisorders/) in the closed groups based in the United States, United Kingdom and Germany upon consultation with their administrators. The survey was online from 12/27/2015 to 01/17/2016 and participation was on voluntary basis.
pnes.data <- read.delim("/Users/Lena/Desktop/daten_pnes.csv", header = TRUE)I had to change the names of the columns to further work with the dataset:
names(pnes.data)[names(pnes.data) == "SD01"] <- "gender"
names(pnes.data)[names(pnes.data) == "SD02_01"] <- "age"
names(pnes.data)[names(pnes.data) == "SD08"] <- "country"
names(pnes.data)[names(pnes.data) == "PS02"] <- "diagnosis"
names(pnes.data)[names(pnes.data) == "PS10"] <- "v.eeg"
names(pnes.data)[names(pnes.data) == "PS04"] <- "latency"
names(pnes.data)[names(pnes.data) == "PS11_01"] <- "trust"
names(pnes.data)[names(pnes.data) == "PS01"] <- "duration"
names(pnes.data)[names(pnes.data) == "PS08"] <- "medication"
names(pnes.data)[names(pnes.data) == "TH01_03"] <- "depression"I eliminated columns that had no further use for the analysis (Section III), that only the following columns remained:
pnes.data <- pnes.data[c("CASE", "gender", "age", "country", "diagnosis", "v.eeg", "latency", "trust", "duration", "medication", "QL01_01", "QL01_02", "QL01_03", "QL01_04", "QL01_05", "QL01_06", "QL01_07", "DD01_01", "DD01_02", "DD01_03", "DD01_04", "DD01_05", "DD01_06", "DD01_07", "DD01_08", "DD01_09", "DD01_10", "TIME_SUM", "depression")]
names(pnes.data)## [1] "CASE" "gender" "age" "country" "diagnosis"
## [6] "v.eeg" "latency" "trust" "duration" "medication"
## [11] "QL01_01" "QL01_02" "QL01_03" "QL01_04" "QL01_05"
## [16] "QL01_06" "QL01_07" "DD01_01" "DD01_02" "DD01_03"
## [21] "DD01_04" "DD01_05" "DD01_06" "DD01_07" "DD01_08"
## [26] "DD01_09" "DD01_10" "TIME_SUM" "depression"Original Columns
- CASE: Participant’s case number
- gender: 1 = female, 2 = male, X-9 = No answer
- age: Participant’s age (in years)
- country: ISO code of the country the participants currently are living in
- diagnosis: 1 = diagnosed with PNES, 2 = not diagnosed wih PNES
- v.eeg: 1 = video EEG monitoring was used for diagnosis 2 = v-EEG monitoring wasn’t used
- latency: of PNES diagnosis (1 = first diagnosis, 2 = some months, 3 = 1 year, 4 = 2 years,… 12 = >/= 10 years)
- trust: into medical workers (1= low to 5 = high)
- duration: of suffering from PNES (1 = < 6 months, 2 = 1 year, 3 = 2 years,…11 = >/= 10 years)
- medication: Ingestion of antiepileptic drugs (AEDs) (1 = yes in the past, 2 = yes in the present, 3 = no, never)
- TIME_SUM: time (in sec.) participants needed to complete the survey
- depression: as a comorbid disease (FALSE = no depression, TRUE = depression)
New Columns
pwi: Personal Wellbeing Index average sum score, measures Subjective Well-being (SWB) through 7 items resp. quality of life domains (old columns): QL01_01, QL01_02, QL01_03,…QL01_07, each ranging from 0 (low satisfaction) to 10 (high satisfaction) (Scale Max. 100%).
desc: Rasch-based Depression Screening sum score, measures the development and severity of depression through 10 questions/items (old columns): DD01_01, DD01_02, DD01_03,…DD01_10, each ranging from 0 (no expression of depression symptom) to 4 (high expression of depression symptom)
medication.conv: Ingestion of AEDs (0 = no, never, 1 = yes, in the past or in the present)
Before starting analysis, I had to eliminate participants who reported “not being diagnosed with PNES”.
pnes.data <- pnes.data[pnes.data$diagnosis == "1", ]The dataset consisted out of 115 rows and 29 columns, in the course of analysis I generated further columns:
nrow(pnes.data)## [1] 115ncol(pnes.data)## [1] 29dim(pnes.data)## [1] 115 29Section II. Questions
- How much time (in minutes) did it take participants to complete the survey?
- What was the mean age and percentage of men and women participating in the survey?
- From which countries were the participants and how was the geographic distribution?
- Do PNES patients still have trust into medical workers like nurses, neurologists or psychotherapists?
- Was there an influence of diagnostic characteristics (latency of the right differential diagnosis, the ingestion of antiepileptic drugs and the use of v EEG monitoring) on participant’s trust into health care providers?
- Is there a difference in the duration of suffering from PNES for participants who take/took or never ingested antiepileptic drugs?
- According to the theory of SWB homeostasis, depression is the loss of Subjective Well-being based on an inoperative cognitive homeostasis mechanism because of the encumbrance of chronic negative living conditions, like suffering from PNES. The question that arises at this point is, if there is a relationship between Subjective Well-being and the development of depression in PNES patients?
- Is the duration of having PNES associated with patient’s Subjective Well-being (major depressiv vs. healthy patients)?
Section III. Analyses
TASK 2 How much time (in minutes) did it take participants to complete the survey?
#recode time from seconds into minutes
pnes.data$TIME_SUM <- (pnes.data$TIME_SUM / 60)
### TASK 2 calculating the mean and standard deviation for the time participants needed to complete the survey
mean(pnes.data$TIME_SUM) ## [1] 8.305652#The mean time amounted to 8.3 minutes
sd(pnes.data$TIME_SUM) ## [1] 2.872186#The standard deviation of time amounted to 2.9 minutesTASK 9 What was the mean age and percentage of men and women participating in the survey?
mean(pnes.data$age)## [1] 28.8#The mean age amounted to 28.8 years
### TASK 9 summary statistics: calculating the mean and standard deviation of age for each manifestion of gender with aggregate
aggregate(age ~ gender, data = pnes.data, FUN = mean) ## gender age
## 1 1 27.90805
## 2 2 31.51852
## 3 X-9 33.00000#The mean age for women amounted to 27.9 years
#The mean age for men amounted to 31.5 years
aggregate(age ~ gender, data = pnes.data, FUN = sd)## gender age
## 1 1 8.673177
## 2 2 6.863338
## 3 X-9 NA#The standard deviation of the age for women amounted to 8.7 years
#The standard deviation of the age for men amounted to 6.9 years
table(pnes.data$gender) / 115 * 100##
## 1 2 X-9
## 75.6521739 23.4782609 0.8695652#76% women and 23% men took part in the survey TASK 10 & 11 From which countries were the participants and how was the geographic distribution?
table(pnes.data$country)##
## 36 40 76 124 276 372 484 710 756 826 840
## 7 1 1 10 20 2 1 1 2 25 45#install.packages("ISOcodes", depend = TRUE)
library("ISOcodes")
data("ISO_3166_1") #I installed a package that contains a dataset (ISO_3166_1) including the ISO codes of every country ("Numeric") as well as the abbrevation of each country ("Alpha_3")
landcode <- ISO_3166_1
names(landcode)[names(landcode) == "Alpha_3"] <- "shortcut" #changing columns into more logical names
names(landcode)[names(landcode) == "Numeric"] <- "code"
#changing "code" to a numeric variable as "country" is
landcode$code <- as.numeric(landcode$code)
### TASK 11 using a loop to change the ISO code into the abbrevation (shortcut) of every country the participants came from
for(i in 1:length(pnes.data$country)) {
pnes.data$country[i] <- landcode$shortcut[which((pnes.data$country[i]) == (landcode$code))]
}
table(pnes.data$country)##
## AUS AUT BRA CAN CHE DEU GBR IRL MEX USA ZAF
## 7 1 1 10 2 20 25 2 1 45 1### TASK 10 using a function to calculate the percentage for each country, how many participants came from there
prop.country <- function(peach) {
prop <- mean(peach) * 100
return(prop)
}
prop.country(pnes.data$country == "AUS")## [1] 6.086957prop.country(pnes.data$country == "AUT")## [1] 0.8695652prop.country(pnes.data$country == "BRA")## [1] 0.8695652prop.country(pnes.data$country == "CAN")## [1] 8.695652prop.country(pnes.data$country == "CHE")## [1] 1.73913prop.country(pnes.data$country == "DEU")## [1] 17.3913prop.country(pnes.data$country == "GBR")## [1] 21.73913prop.country(pnes.data$country == "IRL")## [1] 1.73913prop.country(pnes.data$country == "MEX")## [1] 0.8695652prop.country(pnes.data$country == "USA")## [1] 39.13043prop.country(pnes.data$country == "ZAF")## [1] 0.8695652Participants came from all over the world: USA (39%), UK (22%), Germany (17%), Canada (9%), Australia (6%), Ireland (2%), Switzerland (2%), Austria (1%), Brazil (1%), Mexico (1%) and South Africa (1%) (percentage values rounded)
TASK 2 & 7 Do PNES patients have still have trust into medical workers like nurses, neurologists or psychotherapists?
### TASK 2 calculating the mean and median of patient's trust
trust.me <- mean(pnes.data$trust)
show(trust.me)## [1] 2.2#In the mean patients reported a trust of 2.2, indicating a low level of trust
trust.med <- median(pnes.data$trust)
show(trust.med)## [1] 2#The median of patient's trust amounted to 2, indicating a low level of trust
### TASK 7 create a histogram for the distribution of patient's trust into medical workers
hist(pnes.data$trust,
main = "Trust into medical workers",
xlab = "Trust (1 = low, 5 = high)",
ylab = "Number of Patients",
ylim = c(0, 50))
abline(v = trust.me, lwd = 2, lty = 1, col = "red") #include line for the mean
text(x = 2.6, y = 45,
labels = paste("Mean = ", round(mean(pnes.data$trust),2), sep = ""), col = "red" )
abline(v = trust.med, lwd = 2, lty = 3, col = "blue")
text(x = 1.65, y = 45, labels = paste("Median = ", round(median(pnes.data$trust), 2), sep = ""), col ="blue" )
TASK 5 & 10 Was there an influence of diagnostic characteristics, latency of the right differential diagnosis, the ingestion of antiepileptic drugs and the use of v EEG monitoring (latency, medication, v.eeg) on participant’s trust into health care providers?
#Using a function to create a new column ("medication.conv") for the distinction regarding the ingestion of AED's to prohibit a positive regr.coefficient
recode.data <- function(data.column, old.data, new.data)
{
new.column <- data.column
for (i in 1:length(data.column))
{
change.log <- new.column == old.data[i]
new.column[change.log] <- new.data[i]
}
return(new.column)
}
pnes.data$medication.conv <- recode.data(pnes.data$medication, c(1,2), c(1,1))
pnes.data$medication.conv <- recode.data(pnes.data$medication.conv, 3, 0)
### TASK 5 Multiple Regression for the influence of the independent variables: latency, medication.conv and v.eeg on the dependent variable: trust
lm.trust <- lm(trust ~ latency + medication.conv + v.eeg, data = pnes.data)
summary(lm.trust)##
## Call:
## lm(formula = trust ~ latency + medication.conv + v.eeg, data = pnes.data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.37613 -0.34927 0.00523 0.47693 3.00523
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.09205 0.31112 9.938 < 2e-16 ***
## latency -0.11817 0.02971 -3.977 0.000125 ***
## medication.conv -1.08151 0.22177 -4.877 3.63e-06 ***
## v.eeg 0.40225 0.18160 2.215 0.028796 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.8321 on 111 degrees of freedom
## Multiple R-squared: 0.5436, Adjusted R-squared: 0.5312
## F-statistic: 44.06 on 3 and 111 DF, p-value: < 2.2e-16###TASK 10 Using a function to correct the estimates into standardized beta values for APA style
stand.beta <- function(estimate, iv, dv) {
stand.estimate <- estimate * (sd(iv) / sd(dv))
return(stand.estimate)
}
stand.beta(lm.trust$coefficients["latency"], pnes.data$latency, pnes.data$trust)## latency
## -0.3499296stand.beta(lm.trust$coefficients["medication.conv"], pnes.data$medication.conv, pnes.data$trust)## medication.conv
## -0.4078423stand.beta(lm.trust$coefficients["v.eeg"], pnes.data$v.eeg, pnes.data$trust)## v.eeg
## 0.1516911Regression analysis indicated, that all three variables showed a significant effect on trust and explained 53% of the variance (R2 = .53, F(3,111) = 44.06, p < .001), whereby the lengths of latency (β = -.35, p < .001) and ingestion of antiepileptic drugs (β = -.41, p < .001) had a significant negative influence in comparison to the use of video EEG monitoring (β = .15, p < .05) having a positive influence on patient’s trust
TASK 1, 3 & 8 Is there a difference in the duration of suffering from PNES for participants who take/took or never ingested antiepileptic drugs?
### TASK 1 recode the variable "medication" into a Stringvariable
pnes.data$medication[pnes.data$medication == 1] <- "Now"
pnes.data$medication[pnes.data$medication == 2] <- "Past"
pnes.data$medication[pnes.data$medication == 3] <- "Never"
### TASK 8 create a boxplot to show the distribution of the duration of having PNES for the different types of AED ingestion
with(pnes.data, boxplot(duration ~ medication, ylab = "duration of having PNES (years)", xlab = "Ingestion of Antiepileptic Drugs (AED)"))
### TASK 3 two sample t-test to test differences in the duration for "AEDs takers" vs. "not AEDs takers"
t.test(duration ~ medication.conv, data = pnes.data)##
## Welch Two Sample t-test
##
## data: duration by medication.conv
## t = -11.042, df = 112.28, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -5.472310 -3.807283
## sample estimates:
## mean in group 0 mean in group 1
## 2.705882 7.345679Patients who never took AEDs reported a sign. lower duration of suffering from PNES (M = 2.71), , in comparison to patients who took or take AEDs (M = 7.35), t(112.28) = -11.04, p < .001.
TASK 4 According to the theory of SWB homeostasis, depression is the loss of Subjective Well-being based on an inoperative cognitive homeostasis mechanism because of the encumbrance of chronic negative living conditions, like suffering from PNES. The question that arises at this point is, if there is a relationship between Subjective Well-being and the development of depression in PNES patients?
#creating a new variable, the average sum score (%) for SWB (pwi), whereby the original scales range from 0 to 10, and the ones in the survey ranged from 1 to 11
pnes.data$pwi <- (((pnes.data$QL01_01 + pnes.data$QL01_02 + pnes.data$QL01_03 + pnes.data$QL01_04 + pnes.data$QL01_05 + pnes.data$QL01_06 + pnes.data$QL01_07) - 7) / 7) * 10
#creating a new variable, the sum score of the depression screening, whereby the original scales range from 0 to 4, and the ones in the survey ranged from 1 to 5
pnes.data$desc <- (pnes.data$DD01_01 + pnes.data$DD01_02 + pnes.data$DD01_03 + pnes.data$DD01_04 + pnes.data$DD01_05 + pnes.data$DD01_06 + pnes.data$DD01_07 + pnes.data$DD01_08 + pnes.data$DD01_09 + pnes.data$DD01_10) - 10
### TASK 4 correlation between SWB (pwi sum score) & the Depression Screening score
cor.test(~ pwi + desc, data = pnes.data)##
## Pearson's product-moment correlation
##
## data: pwi and desc
## t = -13.572, df = 113, p-value < 2.2e-16
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## -0.8481183 -0.7059242
## sample estimates:
## cor
## -0.7872704plot(x = pnes.data$pwi, y = pnes.data$desc, main = "Correlation: SWB & Depression", ylab = "DESC sum score", xlab = "PWI sum score", col = "turquoise3")
abline(lm(desc ~ pwi, data = pnes.data), col = "blue")
pnes.data$depression <- as.numeric(pnes.data$depression)
### TASK 4 correlation between SWB (pwi sum score) & Depression as a reported comorbid disorder
cor.test(~ pwi + depression, data = pnes.data)##
## Pearson's product-moment correlation
##
## data: pwi and depression
## t = -2.0426, df = 113, p-value = 0.04342
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## -0.35939155 -0.00578859
## sample estimates:
## cor
## -0.1886991There was a strong significant negative correlation between the SWB of patients and the development of depression, r(113) = -.79, p < .001. In comparison the reported comorbid disorder of depression and the SWB of patients was only weakly correlated r(113) = -.19, p < .05. Those results underline the concept of the theory of homeostasis, whereby the development of depression is a result of suffering from PNES for years.
TASK 6 Is the duration of having PNES associated with patient’s Subjective Well-being (major depressiv vs. healthy patients)?
### TASK 6 creating a scatterplot for the association between the duration of having PNES and patient's SWB (major depressiv vs. healthy)
plot(1,
xlim = c(0, 10),
ylim = c(0, 90),
type = "n",
main = "Patient's duration of having PNES & Subjective Wellbeing",
xlab = "Duration",
ylab = "Subjective Well-being"
)
#adding lines to the plot
abline(h = seq(0, 100, 5),
lwd = .5,
col = gray(.8)
)
abline(v = seq(0, 100, 5),
lwd = .5,
col = gray(.8)
)
#adding the regression line (influence duration on SWB)
abline(lm(pwi ~ duration, data = pnes.data), col = "blue")
#adding points to the plot
with(subset(pnes.data, desc > 20),
points(duration,
pwi,
pch = 16, col = "red"
))
with(subset(pnes.data, desc < 20),
points(duration,
pwi,
pch = 16, col = "blue"
))
#adding a legend to the color of the points
legend("topright",
c("depressiv", "healthy"),
pch = 16,
col = c("red", "blue")
)
#Correlation between the duration of having PNES and the SWB for control
cor.test(~ duration + pwi, data = pnes.data)##
## Pearson's product-moment correlation
##
## data: duration and pwi
## t = -6.2175, df = 113, p-value = 8.731e-09
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## -0.6297631 -0.3545383
## sample estimates:
## cor
## -0.5048726#Correlation between the duration
cor.test(~ duration + desc, data = pnes.data)##
## Pearson's product-moment correlation
##
## data: duration and desc
## t = 5.6679, df = 113, p-value = 1.122e-07
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## 0.3144712 0.6017576
## sample estimates:
## cor
## 0.4704895The duration of having PNES was associated with patient’s Subjective Well-being (major depressiv vs. healthy patients), this can be underlined by the fact that there was a significant negative correlation between the SWB of patients and the duration of having PNES, r(113) = -.50, p < .001. Besides the reported the duration of having PNES and the development of depression significantly correlated positively, r(113) = .47, p < .05. Those results underline the concept of the theory of homeostasis, whereby the development of depression is a result of suffering from PNES for years.
Section IV. Conclusion
115 patients were finally included in the analysis of the PNES Online survey, which took the participants in the mean 8.3 minutes (SD = 2.9 min.) to complete. The mean age of the participants was 28.8 years (women: M = 27.9, SD = 8.7 years; men: M = 31.5 years, SD = 6.9 years), whereby the sex distribution of attending patients amounted to 76% women, reflecting the reported sex distribution of prevalence. Participants came from all over the world: USA (39%), UK (22%), Germany (17%), Canada (9%), Australia (6%), Ireland (2%), Switzerland (2%), Austria (1%), Brazil (1%), Mexico (1%) and South Africa (1%), buttressing the international connectedness of the used web-based support groups.
In the mean participants reported to disagree when it comes to the question of trust into health care providers like nurses, neurologists or psychotherapists (M = 2.2, SD = 1.2, Median = 2) . Most of the PNES patients (37%) even strongly disagreed, when they were asked, if they had the feeling that their medical condition of having psychogenic-non epileptic seizures is taken seriously and well known by medical workers, which speaks for a big leak of trust and reflects the findings about doctor’s attitudes towards PNES patients by Shneker and Elliot (2008). Regression analysis indicated that all diagnosis characteristics (latency, ingestion of AEDs and the use of video EEG monitoring) had a significant influence on participant’s trust into medical workers and therefore explained 53% of the variance (R^2 = .53, F(3, 111) = 44.06, p < .001), whereby the lengths of latency (β = -.35, p < .001) and ingestion of antiepileptic drugs (β = -.41, p < .001) had a significant negative influence in comparison to the use of video EEG monitoring (β = .15, p < .05) having a positive influence on patient’s trust.
Further results revealed that more than half (58%) of the participants took antiepileptic drugs (AEDs) against their non-epileptic seizures in the past, 12% still did, and only 30% never took antiepileptic drugs. Whereby Patients who never took AEDs reported a sign. lower duration of suffering from PNES (M = 2.71), in comparison to patients who took or take AEDs (M = 7.35), t(112.28) = -11.04, p < .001.
Cummins (1995, 1998, 2000a) suggests in the theory of SWB homeostasis that there is a dynamic biological mechanism regulating SWB to preserve a positive view of life and that clinical depression is the loss of SWB, as a result of an inoperative homeostatic defense system. According to these findings, a chronic negative life condition - like suffering from PNES for years - challenges the cognitive homeostatic control and is likely to fail, which finds expression in depression development (Cummins, Lau & Davern, 2007). The results buttressed this theory by examining a strong significant negative correlation between the SWB of patients and the development of depression, r(113) = -.79, p < .001. In comparison the reported comorbid disorder of depression and the SWB of patients was only weakly correlated r(113) = -.19, p < .05, which speaks for the development of depression as a result of suffering from PNES over years.
Besides results showed that the duration of having PNES is associated with patient’s Subjective Well-being (major depressiv vs. healthy patients): A significant negative correlation between the SWB of patients and the duration of having PNES, r(113) = -.50, p < .001 could be shown. Further the reported duration of having PNES and the development of depression significantly correlated positively, r(113) = .47, p < .05. Also those results underline the concept of the theory of homeostasis.