Echo Analysis
Doc Larry Fulton
7 November 2020
Exploratory Data Analysis
- Methods. Chagas Disease training was provided in an Echo session to two different primary populations of interest, CHW and physicians. These populations completed a pre-examination, a post-examination, and a follow-up assessment of the training. The training was given on the following days. This study was approved by Institutional Review Board protocol number XXXX on XXXXX. Participants were provided informed consent. A copy of the survey is available as Appendix A.
Load libraries and data
#####################Read and Pre-Clean the Data#######################
require(Amelia)## Loading required package: Amelia## Loading required package: Rcpp## ##
## ## Amelia II: Multiple Imputation
## ## (Version 1.7.6, built: 2019-11-24)
## ## Copyright (C) 2005-2021 James Honaker, Gary King and Matthew Blackwell
## ## Refer to http://gking.harvard.edu/amelia/ for more information
## ##require(car)## Loading required package: car## Loading required package: carDatarequire(carData)
require(corrplot)## Loading required package: corrplot## corrplot 0.84 loadedrequire(ggcorrplot)## Loading required package: ggcorrplot## Loading required package: ggplot2require(ggpubr)## Loading required package: ggpubrrequire(gridExtra)## Loading required package: gridExtrarequire(gtable)## Loading required package: gtablerequire(heplots)## Loading required package: heplotsrequire(kableExtra)## Loading required package: kableExtrarequire(leaps)## Loading required package: leapsrequire(MANOVA.RM)## Loading required package: MANOVA.RMrequire(MASS)## Loading required package: MASSrequire(MVN)## Loading required package: MVN## Registered S3 method overwritten by 'GGally':
## method from
## +.gg ggplot2## sROC 0.1-2 loadedrequire(mvtnorm)## Loading required package: mvtnormrequire(nnet)## Loading required package: nnetrequire(psych) #to describe## Loading required package: psych##
## Attaching package: 'psych'## The following objects are masked from 'package:ggplot2':
##
## %+%, alpha## The following object is masked from 'package:car':
##
## logitrequire(ggplot2)
require(ggcorrplot)
require(qcc)## Loading required package: qcc## Package 'qcc' version 2.7## Type 'citation("qcc")' for citing this R package in publications.require(rcompanion)## Loading required package: rcompanion## Error: package or namespace load failed for 'rcompanion' in loadNamespace(j <- i[[1L]], c(lib.loc, .libPaths()), versionCheck = vI[[j]]):
## there is no package called 'coin'require(reticulate) #to use Python in R as well## Loading required package: reticulaterequire(reshape2)## Loading required package: reshape2require(ResourceSelection)## Loading required package: ResourceSelection## ResourceSelection 0.3-5 2019-07-22require(rstatix)## Loading required package: rstatix##
## Attaching package: 'rstatix'## The following object is masked from 'package:MASS':
##
## select## The following object is masked from 'package:ggcorrplot':
##
## cor_pmat## The following object is masked from 'package:stats':
##
## filterrequire(tidyverse)## Loading required package: tidyverse## -- Attaching packages --------------------------------------- tidyverse 1.3.0 --## v tibble 3.0.6 v dplyr 1.0.4
## v tidyr 1.1.2 v stringr 1.4.0
## v readr 1.4.0 v forcats 0.5.1
## v purrr 0.3.4## Warning: package 'stringr' was built under R version 4.0.4## -- Conflicts ------------------------------------------ tidyverse_conflicts() --
## x psych::%+%() masks ggplot2::%+%()
## x psych::alpha() masks ggplot2::alpha()
## x dplyr::combine() masks gridExtra::combine()
## x dplyr::filter() masks rstatix::filter(), stats::filter()
## x dplyr::group_rows() masks kableExtra::group_rows()
## x dplyr::lag() masks stats::lag()
## x dplyr::recode() masks car::recode()
## x dplyr::select() masks rstatix::select(), MASS::select()
## x purrr::some() masks car::some()corfunction=function(d){
mycorr=cor(d[, 1:ncol(d)]); p.mat=ggcorrplot::cor_pmat(d[,1:ncol(d)])
myplot=ggcorrplot(mycorr, hc.order=TRUE,type="lower",colors=c("red", "white","green"),tl.cex = 8, tl.col = "black", lab=TRUE, lab_size=2, p.mat=p.mat, insig="pch", pch=4)
print(myplot)}
#Barplot Function, Vertical
mybar2=function(y,title){
fu%>%count(fu[,y])%>%mutate(perc = n / nrow(fu)) -> mydf1
a=ggplot(mydf1, aes(x='',n, fill=`fu[, y]`,label=paste0(n," (", scales::percent(perc), ")")))+
geom_bar(position="stack", stat="identity", width=.5)+
geom_text(size = 3, angle=90, position = position_stack(vjust = 0.5))+
xlab("")+
ylab("")+
coord_flip()+
theme(legend.title = element_blank())+
ggtitle(title)+
theme(legend.text = element_text(size = 8))+
theme(legend.position="right")
return(a)
}
mybar3=function(y,varname,rotate=270){
myt=table(y, mydata$PrePost)
myt=apply(myt,2,function(z) z/sum(z))
myt=as.data.frame(myt)
myt$Prepost=rownames(myt)
myt=melt(myt, id.vars="Prepost")
colnames(myt)=c("Prepost", varname,"Percent")
a=ggplot(myt, aes(myt[,2], Percent, fill=Prepost, label=scales::percent(Percent)))+
geom_bar( stat = "identity", position = "stack" ) +
geom_text(size = 3, angle=rotate,position = position_stack(vjust = 0.5))+
coord_flip() +
xlab("")+
ylab("")+
theme_minimal() + theme(legend.title=element_blank(),legend.position = "bottom" )+
ggtitle(varname)
return(a)
}
setwd("D:/PaulaEcho")
mydata=read.csv("global.csv", stringsAsFactors = TRUE)
#########################################################################Check Missing
3.1. Missing Data.
After eliminating those who did not fully complete the pre-post examination or follow-ups or who opted out during informed consent, there were n1=183 pretests and n2=71 post tests. For the follow-up survey, there were 54 completed surveys. Due to optional anonymity, only n3=17 pretests were matched with post-tests. The available data were complete.
#########################################################################
missmap(mydata)
#########################################################################Descriptives
- Results
4.1. Descriptive Statistics
Figure 1 provides the distribution of the gender, ethnicity, age, and profession of both the pre-post groups. This information was not gathered for the follow-up survey. About 74% of the pre-test participants were female as well as 85% of the post-test participants. Tha majority of both of these populations were Hispanic (55% pre-test, 70% post-test), and the age distribution was modally 45-64 for both groups. The majority of the individuals completing the pre-test were MD’s or other (32% each), while the majority completing the post-test were CHWs (56%).
Gender
#########################################################################
pre=mydata[mydata$PrePost=="Pre",]
post=mydata[mydata$PrePost=="Post",]
as.data.frame(100*round(table(mydata$Gender)/length(mydata$Gender),4))%>%
kbl(col.names = c("Gender", "%"))%>%kable_classic(full_width=F)| Gender | % |
|---|---|
| Female | 77.17 |
| Male | 22.83 |
as.data.frame(round(table(mydata$Gender),4))%>%
kbl(col.names = c("Gender", "n"))%>%kable_classic(full_width=F)| Gender | n |
|---|---|
| Female | 196 |
| Male | 58 |
as.data.frame(round(table(pre$Gender),4))%>%
kbl(col.names = c("Pre Gender", "n"))%>%kable_classic(full_width=F)| Pre Gender | n |
|---|---|
| Female | 136 |
| Male | 47 |
as.data.frame(round(table(post$Gender),4))%>%
kbl(col.names = c("Post Gender", "n"))%>%kable_classic(full_width=F)| Post Gender | n |
|---|---|
| Female | 60 |
| Male | 11 |
mybar3(mydata$Gender,"Gender")
#########################################################################Ethnicity
#########################################################################
as.data.frame(100*round(table(mydata$Ethnicity)/length(mydata$Ethnicity),4))%>% kbl(col.names = c("Ethnicity", "%"))%>%kable_classic(full_width=F)| Ethnicity | % |
|---|---|
| Hispanic | 58.27 |
| Non-Hispanic | 38.19 |
| Non -Hispanic | 0.39 |
| Prefer not to respond | 3.15 |
as.data.frame(round(table(mydata$Ethnicity),4))%>% kbl(col.names = c("Ethnicity", "n"))%>%kable_classic(full_width=F)| Ethnicity | n |
|---|---|
| Hispanic | 148 |
| Non-Hispanic | 97 |
| Non -Hispanic | 1 |
| Prefer not to respond | 8 |
as.data.frame(round(table(pre$Ethnicity),4))%>% kbl(col.names = c("Pre Ethnicity", "n"))%>%kable_classic(full_width=F)| Pre Ethnicity | n |
|---|---|
| Hispanic | 98 |
| Non-Hispanic | 78 |
| Non -Hispanic | 1 |
| Prefer not to respond | 6 |
as.data.frame(round(table(post$Ethnicity),4))%>% kbl(col.names = c("Post Ethnicity", "n"))%>%kable_classic(full_width=F)| Post Ethnicity | n |
|---|---|
| Hispanic | 50 |
| Non-Hispanic | 19 |
| Non -Hispanic | 0 |
| Prefer not to respond | 2 |
mybar3(mydata$Ethnicity,"Ethnicity")
#########################################################################Age
#########################################################################
as.data.frame(100*round(table(mydata$Age)/length(mydata$Age),4))%>%
kbl(col.names = c("Age", "%"))%>%kable_classic(full_width=F)| Age | % |
|---|---|
| 18 to 24 | 5.51 |
| 25 to 34 | 18.90 |
| 35 to 44 | 26.38 |
| 45 to 64 | 41.73 |
| 65 and over | 6.30 |
| Prefer not to respond | 1.18 |
as.data.frame(round(table(mydata$Age),4))%>%
kbl(col.names = c("Age", "n"))%>%kable_classic(full_width=F)| Age | n |
|---|---|
| 18 to 24 | 14 |
| 25 to 34 | 48 |
| 35 to 44 | 67 |
| 45 to 64 | 106 |
| 65 and over | 16 |
| Prefer not to respond | 3 |
as.data.frame(round(table(pre$Age),4))%>%
kbl(col.names = c("Pre Age", "n"))%>%kable_classic(full_width=F)| Pre Age | n |
|---|---|
| 18 to 24 | 12 |
| 25 to 34 | 38 |
| 35 to 44 | 54 |
| 45 to 64 | 68 |
| 65 and over | 10 |
| Prefer not to respond | 1 |
as.data.frame(round(table(post$Age),4))%>%
kbl(col.names = c("Post Age", "n"))%>%kable_classic(full_width=F)| Post Age | n |
|---|---|
| 18 to 24 | 2 |
| 25 to 34 | 10 |
| 35 to 44 | 13 |
| 45 to 64 | 38 |
| 65 and over | 6 |
| Prefer not to respond | 2 |
mybar3(mydata$Age,"Age")
#########################################################################Profession
#########################################################################
as.data.frame(100*round(table(mydata$Profession)/length(mydata$Profession),4))%>%
kbl(col.names = c("Profession", "%"))%>%kable_classic(full_width=F)| Profession | % |
|---|---|
| CHW | 37.01 |
| DVM | 3.54 |
| MD/DO | 28.74 |
| NP/PA | 2.36 |
| Other | 28.35 |
as.data.frame(round(table(mydata$Profession),4))%>%
kbl(col.names = c("Profession", "n"))%>%kable_classic(full_width=F)| Profession | n |
|---|---|
| CHW | 94 |
| DVM | 9 |
| MD/DO | 73 |
| NP/PA | 6 |
| Other | 72 |
as.data.frame(round(table(pre$Profession),4))%>%
kbl(col.names = c("Pre Profession", "n"))%>%kable_classic(full_width=F)| Pre Profession | n |
|---|---|
| CHW | 54 |
| DVM | 7 |
| MD/DO | 58 |
| NP/PA | 6 |
| Other | 58 |
as.data.frame(round(table(post$Profession),4))%>%
kbl(col.names = c("Post Profession", "n"))%>%kable_classic(full_width=F)| Post Profession | n |
|---|---|
| CHW | 40 |
| DVM | 2 |
| MD/DO | 15 |
| NP/PA | 0 |
| Other | 14 |
mybar3(mydata$Profession,"Profession")
#########################################################################Specialization
#########################################################################
as.data.frame(100*round(table(mydata$Specialization)/length(mydata$Specialization),4))%>%
kbl(col.names = c("Specialization", "%"))%>%kable_classic(full_width=F)| Specialization | % |
|---|---|
| 42.91 | |
| Air Force idmt | 0.39 |
| Air Force IDMT | 0.39 |
| Biomedicine | 0.39 |
| Community health | 28.35 |
| Community health,Other | 0.39 |
| DC | 0.39 |
| Does not apply | 2.76 |
| economist | 0.39 |
| Economist | 0.39 |
| Epidemiologist | 1.18 |
| FSC | 0.39 |
| Infectious disease,Cardiology,General practitioner/Family physician,Community health | 0.79 |
| LVN | 0.39 |
| Microbiologist | 0.39 |
| MLS | 0.39 |
| Other | 1.18 |
| Other / Does not apply | 4.33 |
| Other,Does not apply | 0.39 |
| Ph.D. / Post-Grad / Faculty | 4.33 |
| PharmD / MPharm / Pharm Student | 4.72 |
| PT, MHA, PhD | 0.39 |
| Public Health | 1.57 |
| R&D Industry | 0.39 |
| RCS | 0.39 |
| Student | 1.57 |
| Zoonosis Control Program Specialist | 0.39 |
as.data.frame(table(mydata$Specialization))%>%
kbl(col.names = c("Specialization", "n"))%>%kable_classic(full_width=F)| Specialization | n |
|---|---|
| 109 | |
| Air Force idmt | 1 |
| Air Force IDMT | 1 |
| Biomedicine | 1 |
| Community health | 72 |
| Community health,Other | 1 |
| DC | 1 |
| Does not apply | 7 |
| economist | 1 |
| Economist | 1 |
| Epidemiologist | 3 |
| FSC | 1 |
| Infectious disease,Cardiology,General practitioner/Family physician,Community health | 2 |
| LVN | 1 |
| Microbiologist | 1 |
| MLS | 1 |
| Other | 3 |
| Other / Does not apply | 11 |
| Other,Does not apply | 1 |
| Ph.D. / Post-Grad / Faculty | 11 |
| PharmD / MPharm / Pharm Student | 12 |
| PT, MHA, PhD | 1 |
| Public Health | 4 |
| R&D Industry | 1 |
| RCS | 1 |
| Student | 4 |
| Zoonosis Control Program Specialist | 1 |
as.data.frame(table(pre$Specialization))%>%
kbl(col.names = c("Pre Specialization", "n"))%>%kable_classic(full_width=F)| Pre Specialization | n |
|---|---|
| 89 | |
| Air Force idmt | 0 |
| Air Force IDMT | 1 |
| Biomedicine | 1 |
| Community health | 39 |
| Community health,Other | 1 |
| DC | 1 |
| Does not apply | 5 |
| economist | 0 |
| Economist | 1 |
| Epidemiologist | 2 |
| FSC | 1 |
| Infectious disease,Cardiology,General practitioner/Family physician,Community health | 0 |
| LVN | 1 |
| Microbiologist | 1 |
| MLS | 1 |
| Other | 3 |
| Other / Does not apply | 8 |
| Other,Does not apply | 0 |
| Ph.D. / Post-Grad / Faculty | 9 |
| PharmD / MPharm / Pharm Student | 10 |
| PT, MHA, PhD | 1 |
| Public Health | 3 |
| R&D Industry | 1 |
| RCS | 1 |
| Student | 2 |
| Zoonosis Control Program Specialist | 1 |
as.data.frame(table(post$Specialization))%>%
kbl(col.names = c("Post Specialization", "n"))%>%kable_classic(full_width=F)| Post Specialization | n |
|---|---|
| 20 | |
| Air Force idmt | 1 |
| Air Force IDMT | 0 |
| Biomedicine | 0 |
| Community health | 33 |
| Community health,Other | 0 |
| DC | 0 |
| Does not apply | 2 |
| economist | 1 |
| Economist | 0 |
| Epidemiologist | 1 |
| FSC | 0 |
| Infectious disease,Cardiology,General practitioner/Family physician,Community health | 2 |
| LVN | 0 |
| Microbiologist | 0 |
| MLS | 0 |
| Other | 0 |
| Other / Does not apply | 3 |
| Other,Does not apply | 1 |
| Ph.D. / Post-Grad / Faculty | 2 |
| PharmD / MPharm / Pharm Student | 2 |
| PT, MHA, PhD | 0 |
| Public Health | 1 |
| R&D Industry | 0 |
| RCS | 0 |
| Student | 2 |
| Zoonosis Control Program Specialist | 0 |
#########################################################################For the pre-test, the modal response when asked about their knowledge of CD was “Good” (30%) with only 8% professing “Excellent” knowledge. After training, that self-assessment increased to 58% “Good” and 16% Excellent (Fisher’s Exact Test p-value<0.001).
Knowledge
#########################################################################
myt=table(mydata$PrePost,mydata$Knowledge)
myt=myt/rowSums(myt)
colnames(myt)=c("None", "Very Limited", "Limited", "Good", "Excellent")
row.names(myt)=c("Post", "Pre")
myt=round(100*myt, 4)
myt%>%
kbl()%>%kable_classic(full_width=F)| None | Very Limited | Limited | Good | Excellent | |
|---|---|---|---|---|---|
| Post | 0.0000 | 5.6338 | 21.1268 | 57.7465 | 15.4930 |
| Pre | 12.5683 | 19.1257 | 32.2404 | 30.0546 | 6.0109 |
mybar3(mydata$Knowledge,"Knowledge")
fisher.test(myt, simulate.p.value = TRUE)## Warning in fisher.test(myt, simulate.p.value = TRUE): 'x' has been rounded to
## integer: Mean relative difference: 0.010514##
## Fisher's Exact Test for Count Data with simulated p-value (based on
## 2000 replicates)
##
## data: myt
## p-value = 0.0004998
## alternative hypothesis: two.sided#########################################################################Prior to the training 75% were not confident or only somewhat confident in the recency of their knowledge. Post-training, 80% were either confident or very confident (Fisher’s Exact Test p-value < .001).
Recency
#########################################################################
myt=table(mydata$PrePost,mydata$Recent)
myt=myt/rowSums(myt)
colnames(myt)=c("Not at all Confident", "Somewhat Confident", "Confident", "Very Confident")
row.names(myt)=c("Post","Pre")
myt=round(100*myt, 4)
myt%>%kbl()%>%kable_classic(full_width=F)| Not at all Confident | Somewhat Confident | Confident | Very Confident | |
|---|---|---|---|---|
| Post | 1.4085 | 16.9014 | 54.9296 | 26.7606 |
| Pre | 40.9836 | 34.4262 | 17.4863 | 7.1038 |
mybar3(mydata$Recent, "Recency of Knowledge")
fisher.test(myt, simulate.p.value = TRUE)## Warning in fisher.test(myt, simulate.p.value = TRUE): 'x' has been rounded to
## integer: Mean relative difference: 0.009248##
## Fisher's Exact Test for Count Data with simulated p-value (based on
## 2000 replicates)
##
## data: myt
## p-value = 0.0004998
## alternative hypothesis: two.sided#########################################################################For the pre-post, students were provided 10 questions to test their knowledge. The distribution of the pre-post scores is shown in Figure 2 (unmatched) with notched boxplots, boxplots that provide a visual median test. Without controlling for individual ability, median post-scores were statistically better than pre-test scores at the alpha=0.05 level.
Score Pre-Post
#########################################################################
mydata$PrePost2=mydata$PrePost
mydata$PrePost=as.factor(mydata$PrePost)
levels(mydata$PrePost)=c("Post", "Pre")
boxplot(mydata$Score~mydata$PrePost, main="Knowledge Score, Pre vs. Post",notch=TRUE, col=c("red", "dark green"), horizontal = TRUE)
#########################################################################For the majority of questions, the pre-post scores appeared to improve (Figure 3). The exception to this statement was for “who should be treated with anti-T drugs” and for “prevention” methods. Both of these topics will be reinforced in future sessions.
Subscores
#########################################################################
mynames=c("True or False: Chagas Disease Present in Texas",
"Chagas Disease Caused By...",
"T Cruzi Transmitted by Saliva.",
"Parts of World for Chagas Disease",
"What % of Patients Develop Clinical Disease",
"Chagas Disease Symptoms...",
"Methods to Confirm Chagas Disease",
"Who should be Treated with Antitrypanosomal Drugs",
"EKG Findings Typical of Chagas Disease",
"Steps for Prevention")
for (i in 10:19){
print(mybar3(mydata[,i],mynames[i-9]))
}
Test questions were found to be uncorrelated, so that performance on one question was statistically not linked to any others. This finding suggests that the topics are sufficiently different.
Correlations
#########################################################################
newdat=mydata[, c(10:19)]
newdat2=pre[, c(10:19)]
newdat3=post[, c(10:19)]
for (i in 1:10){ newdat[,i]=as.numeric(newdat[,i])-1
newdat2[,i]=as.numeric(newdat2[,i])-1
newdat3[,i]=as.numeric(newdat3[,i])-1
}
mycorr=cor(newdat)
mycorr2=cor(newdat2)
mycorr3=cor(newdat3)## Warning in cor(newdat3): the standard deviation is zerocorfunction(mycorr)
corfunction(mycorr2)
#########################################################################Table 1 provides the descriptive statistics for the comparison between the pre-test and post-test. These scores are not matched.
4.2. Inferential Testing
Both paired and unpaired, Holm-adjusted t-tests were used to evaluate performance differences for the students. The unpaired t-tests do not account for individual knowledge coming into the class; however, they allow for a larger sample size.
For the unpaired t-tests (Table 2), we found statistically significant differences for all but three of the questions associated with symptoms, diagnoses, and prevention. These are some of the more difficult concepts taught in the course.
When comparing CHWs with non-CHWs (Table 3), we noticed that the CHW workers performed worse than the non-CHWs (many of whom were doctors) as indicated by the negative difference between the scores. There were no statistically significant differences on 4 of the 10 questions, though.
Simple Inferentials
Paired t-test (Small Subset)
prepost=read.csv("D:/PaulaEcho/prepost.csv", stringsAsFactors = T)
MD=prepost[prepost$Group=="MD",]
CHW=prepost[prepost$Group=="CHW",]
prepost$Group=MD$Group=CHW$Group=NULL
myf2=function(x){
myt=t.test(x)
newp=round(p.adjust(myt$p.value),3)
a=c(round(myt$estimate[1],3),
round(myt$conf.int[1],3),
round(myt$conf.int[2],3),round(myt$statistic,3),
round(myt$parameter,3),round(newp,3))
return(a)
}
a=matrix(rep(NA, 6*12), 12)
for (i in 1:12){a[i,1:6]=myf2(prepost[,i])}
a=as.data.frame(a)
for (i in 1:6){a[,i]=as.numeric(a[,i])}
rownames(a)=colnames(prepost)
colnames(a)=c("Mean","Lower 95% CI", "Upper 95% CI", "t-Value", "df", "Holm-Adjusted p")
a%>%kbl(caption="Pre-Post ALL, n=45")%>%kable_classic()| Mean | Lower 95% CI | Upper 95% CI | t-Value | df | Holm-Adjusted p | |
|---|---|---|---|---|---|---|
| Knowledge | 0.867 | 0.469 | 1.265 | 4.389 | 44 | 0.000 |
| Recent | 1.044 | 0.731 | 1.358 | 6.714 | 44 | 0.000 |
| Texas | 0.022 | -0.023 | 0.067 | 1.000 | 44 | 0.323 |
| Cause | 0.200 | 0.063 | 0.337 | 2.934 | 44 | 0.005 |
| Saliva | 0.289 | 0.112 | 0.466 | 3.292 | 44 | 0.002 |
| Location | 0.067 | -0.033 | 0.166 | 1.354 | 44 | 0.183 |
| Prevalence | 0.356 | 0.173 | 0.538 | 3.917 | 44 | 0.000 |
| ClinDisease | 0.133 | -0.018 | 0.285 | 1.773 | 44 | 0.083 |
| Symptoms | 0.156 | -0.025 | 0.336 | 1.735 | 44 | 0.090 |
| Diagnosis | -0.022 | -0.220 | 0.175 | -0.227 | 44 | 0.821 |
| AntiT_Drugs | 0.111 | -0.049 | 0.271 | 1.402 | 44 | 0.168 |
| Prevent | -0.178 | -0.351 | -0.005 | -2.072 | 44 | 0.044 |
a=matrix(rep(NA, 6*12), 12)
for (i in 1:12){a[i,1:6]=myf2(MD[,i])}
a=as.data.frame(a)
for (i in 1:6){a[,i]=as.numeric(a[,i])}
rownames(a)=colnames(MD)
colnames(a)=c("Mean","Lower 95% CI", "Upper 95% CI", "t-Value", "df", "Holm-Adjusted p")
a%>%kbl(caption="Pre-Post MD, n=28")%>%kable_classic()| Mean | Lower 95% CI | Upper 95% CI | t-Value | df | Holm-Adjusted p | |
|---|---|---|---|---|---|---|
| Knowledge | 0.250 | -0.167 | 0.667 | 1.230 | 27 | 0.229 |
| Recent | 0.643 | 0.259 | 1.026 | 3.438 | 27 | 0.002 |
| Texas | 0.000 | NaN | NaN | NaN | 27 | NaN |
| Cause | 0.071 | -0.075 | 0.218 | 1.000 | 27 | 0.326 |
| Saliva | 0.179 | -0.006 | 0.363 | 1.987 | 27 | 0.057 |
| Location | 0.000 | -0.106 | 0.106 | 0.000 | 27 | 1.000 |
| Prevalence | 0.179 | -0.059 | 0.416 | 1.544 | 27 | 0.134 |
| ClinDisease | -0.071 | -0.218 | 0.075 | -1.000 | 27 | 0.326 |
| Symptoms | 0.000 | -0.236 | 0.236 | 0.000 | 27 | 1.000 |
| Diagnosis | 0.036 | -0.188 | 0.259 | 0.328 | 27 | 0.745 |
| AntiT_Drugs | 0.036 | -0.161 | 0.233 | 0.372 | 27 | 0.713 |
| Prevent | -0.107 | -0.351 | 0.137 | -0.902 | 27 | 0.375 |
a=matrix(rep(NA, 6*12), 12)
for (i in 1:12){a[i,1:6]=myf2(CHW[,i])}
a=as.data.frame(a)
for (i in 1:6){a[,i]=as.numeric(a[,i])}
rownames(a)=colnames(CHW)
colnames(a)=c("Mean","Lower 95% CI", "Upper 95% CI", "t-Value", "df", "Holm-Adjusted p")
a%>%kbl(caption="Pre-Post CHW, n=17")%>%kable_classic()| Mean | Lower 95% CI | Upper 95% CI | t-Value | df | Holm-Adjusted p | |
|---|---|---|---|---|---|---|
| Knowledge | 1.882 | 1.341 | 2.424 | 7.366 | 16 | 0.000 |
| Recent | 1.706 | 1.309 | 2.103 | 9.114 | 16 | 0.000 |
| Texas | 0.059 | -0.066 | 0.184 | 1.000 | 16 | 0.332 |
| Cause | 0.412 | 0.151 | 0.673 | 3.347 | 16 | 0.004 |
| Saliva | 0.471 | 0.102 | 0.839 | 2.704 | 16 | 0.016 |
| Location | 0.176 | -0.026 | 0.379 | 1.852 | 16 | 0.083 |
| Prevalence | 0.647 | 0.394 | 0.900 | 5.416 | 16 | 0.000 |
| ClinDisease | 0.471 | 0.206 | 0.735 | 3.771 | 16 | 0.002 |
| Symptoms | 0.412 | 0.151 | 0.673 | 3.347 | 16 | 0.004 |
| Diagnosis | -0.118 | -0.519 | 0.284 | -0.621 | 16 | 0.543 |
| AntiT_Drugs | 0.235 | -0.054 | 0.524 | 1.725 | 16 | 0.104 |
| Prevent | -0.294 | -0.536 | -0.053 | -2.582 | 16 | 0.020 |
A course satisfaction survey revealed wide-spread approval of the content and instructors. The individual questions are Appendix B, and the results of the survey are in Table 5.
Follow-Up Satisfaction
fu=read.csv("D:/PaulaEcho/total followup.csv")
describe(fu)%>%kbl(caption="CHW Follow-Up")%>%kable_classic(html_font = "Cambria")| vars | n | mean | sd | median | trimmed | mad | min | max | range | skew | kurtosis | se | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Occupation* | 1 | 54 | 2.574074 | 1.6437630 | 3 | 2.318182 | 1.4826 | 1 | 8 | 7 | 1.1847971 | 1.3818618 | 0.2236878 |
| Learning.objectives.were.clearly.stated* | 2 | 54 | 2.555556 | 0.5378738 | 3 | 2.590909 | 0.0000 | 1 | 3 | 2 | -0.5729870 | -0.9732335 | 0.0731953 |
| Content.was.well.organized* | 3 | 54 | 2.574074 | 0.5697352 | 3 | 2.636364 | 0.0000 | 1 | 3 | 2 | -0.8856959 | -0.2884584 | 0.0775311 |
| Stated.learning.objectives.were.met* | 4 | 54 | 1.555556 | 0.5015699 | 2 | 1.568182 | 0.0000 | 1 | 2 | 1 | -0.2174243 | -1.9885288 | 0.0682550 |
| Content.was.accurate.and.relevant* | 5 | 54 | 3.518518 | 0.6656176 | 4 | 3.613636 | 0.0000 | 1 | 4 | 3 | -1.3849570 | 2.0907518 | 0.0905791 |
| Course.corresponded.to.my.expectations* | 6 | 54 | 3.351852 | 0.6488715 | 3 | 3.409091 | 0.0000 | 1 | 4 | 3 | -0.8748751 | 1.3626716 | 0.0883002 |
| I.intend.to.apply.what.I.learned* | 7 | 54 | 2.481482 | 0.6062797 | 3 | 2.545454 | 0.0000 | 1 | 3 | 2 | -0.6786852 | -0.5699189 | 0.0825042 |
| The.lecturers.demonstrated.knowledge.of.the.subject.matter* | 8 | 54 | 1.703704 | 0.4609109 | 2 | 1.750000 | 0.0000 | 1 | 2 | 1 | -0.8675491 | -1.2698519 | 0.0627220 |
| The.lecturers.were.effective.in.communicating.the.content.of.the.training.program* | 9 | 54 | 1.648148 | 0.4820322 | 2 | 1.681818 | 0.0000 | 1 | 2 | 1 | -0.6032986 | -1.6658575 | 0.0655963 |
| The.lecturers.encouraged.feedback.from.the.participants* | 10 | 54 | 2.592593 | 0.5669687 | 3 | 2.659091 | 0.0000 | 1 | 3 | 2 | -0.9678568 | -0.1285472 | 0.0771547 |
| Overall..I.would.rate.this.program* | 11 | 54 | 2.722222 | 0.4920756 | 3 | 2.795454 | 0.0000 | 1 | 3 | 2 | -1.4218180 | 0.9710169 | 0.0669630 |
| Overall..I.would.rate.the.lecturers* | 12 | 54 | 1.777778 | 0.4196435 | 2 | 1.840909 | 0.0000 | 1 | 2 | 1 | -1.2993590 | -0.3165050 | 0.0571062 |
| Did.you.find.this.activity.enhanced.your.understanding.of.Chagas.disease.* | 13 | 54 | 1.981481 | 0.1360828 | 2 | 2.000000 | 0.0000 | 1 | 2 | 1 | -6.9452610 | 47.1100823 | 0.0185185 |
| Would.you.recommend.this.training.program.to.other.clinicians..clinical.staff.members..medical.students..or.community.health.workers.* | 14 | 54 | 1.000000 | 0.0000000 | 1 | 1.000000 | 0.0000 | 1 | 1 | 0 | NaN | NaN | 0.0000000 |
| Was.there.content.you.expected.to.be.covered.that.was.not.* | 15 | 54 | 1.240741 | 0.4315477 | 1 | 1.181818 | 0.0000 | 1 | 2 | 1 | 1.1792818 | -0.6197488 | 0.0587262 |
| If.yes..please.detail.below.* | 16 | 53 | 1.849057 | 2.1785336 | 1 | 1.232558 | 0.0000 | 1 | 10 | 9 | 2.4890049 | 5.0463348 | 0.2992446 |
| Please.provide.additional.comments.and.feedback..if.any..* | 17 | 54 | 3.833333 | 5.0497525 | 1 | 2.772727 | 0.0000 | 1 | 18 | 17 | 1.5377847 | 0.9188453 | 0.6871843 |
fu=fu[, 1:15]
for (i in 2:15){
print(mybar2(i,colnames(fu)[i]))
}
Models
A linear model of performance score as a function of age, ethnicity, gender, self-assessed knowledge, recency of knowledge, CHW status, and pre-post status of examination was estimated using ordinary least squares on the n1+n2=254 tests. The results showed that age groupings, ethnicity, self-assessment of knowledge, and pre-post examination were important in predicting test results (F(13,240)=13.3, p<.001, R^2=.440, Adj. R^2=.410).
Regression
library(lmtest)## Loading required package: zoo##
## Attaching package: 'zoo'## The following objects are masked from 'package:base':
##
## as.Date, as.Date.numericmydata$CHW=mydata$Profession
mydata$CHW=rep(0, nrow(mydata))
mydata$CHW[mydata$Profession=="CHW"]=1
mydata$CHW=as.numeric(mydata$CHW)
mydata$Female=as.numeric(mydata$Gender)
mydata$Female[mydata$Female==2]=0
mylm=lm(Score~Age+Female+Ethnicity+Knowledge+Recent+CHW+PrePost2, data=mydata)
summary(mylm)$coefficients%>%kbl()%>%kable_classic(html_font = 'Cambria')| Estimate | Std. Error | t value | Pr(>|t|) | |
|---|---|---|---|---|
| (Intercept) | 0.4072243 | 0.0636530 | 6.3975652 | 0.0000000 |
| Age25 to 34 | 0.2332343 | 0.0464262 | 5.0237633 | 0.0000010 |
| Age35 to 44 | 0.1639557 | 0.0461276 | 3.5543908 | 0.0004579 |
| Age45 to 64 | 0.1678968 | 0.0445374 | 3.7697960 | 0.0002067 |
| Age65 and over | 0.1799692 | 0.0576812 | 3.1200660 | 0.0020342 |
| AgePrefer not to respond | 0.1781863 | 0.0987719 | 1.8040180 | 0.0725090 |
| Female | -0.0175723 | 0.0248363 | -0.7075240 | 0.4799420 |
| EthnicityNon-Hispanic | 0.0981494 | 0.0271304 | 3.6176931 | 0.0003638 |
| EthnicityNon -Hispanic | -0.0323416 | 0.1509386 | -0.2142698 | 0.8305225 |
| EthnicityPrefer not to respond | -0.0165068 | 0.0593320 | -0.2782100 | 0.7810958 |
| Knowledge2. Very Limited | -0.0171143 | 0.0402922 | -0.4247539 | 0.6714047 |
| Knowledge3. Limited | 0.0790134 | 0.0420758 | 1.8778821 | 0.0616357 |
| Knowledge4. Good | 0.1289828 | 0.0460063 | 2.8035876 | 0.0054760 |
| Knowledge5. Excellent | 0.1441796 | 0.0594176 | 2.4265454 | 0.0159967 |
| Recent2. Somewhat Confident | 0.0195944 | 0.0301687 | 0.6494920 | 0.5166548 |
| Recent3. Confident | 0.0388984 | 0.0395262 | 0.9841180 | 0.3260696 |
| Recent4. Very Confident | 0.0293292 | 0.0498102 | 0.5888186 | 0.5565485 |
| CHW | -0.0592721 | 0.0290590 | -2.0397173 | 0.0424971 |
| PrePost2Pre | -0.0891525 | 0.0266958 | -3.3395752 | 0.0009757 |
summary(mylm)##
## Call:
## lm(formula = Score ~ Age + Female + Ethnicity + Knowledge + Recent +
## CHW + PrePost2, data = mydata)
##
## Residuals:
## Min 1Q Median 3Q Max
## -0.39158 -0.09124 -0.00025 0.09226 0.40799
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.40722 0.06365 6.398 8.46e-10 ***
## Age25 to 34 0.23323 0.04643 5.024 1.00e-06 ***
## Age35 to 44 0.16396 0.04613 3.554 0.000458 ***
## Age45 to 64 0.16790 0.04454 3.770 0.000207 ***
## Age65 and over 0.17997 0.05768 3.120 0.002034 **
## AgePrefer not to respond 0.17819 0.09877 1.804 0.072509 .
## Female -0.01757 0.02484 -0.708 0.479942
## EthnicityNon-Hispanic 0.09815 0.02713 3.618 0.000364 ***
## EthnicityNon -Hispanic -0.03234 0.15094 -0.214 0.830522
## EthnicityPrefer not to respond -0.01651 0.05933 -0.278 0.781096
## Knowledge2. Very Limited -0.01711 0.04029 -0.425 0.671405
## Knowledge3. Limited 0.07901 0.04208 1.878 0.061636 .
## Knowledge4. Good 0.12898 0.04601 2.804 0.005476 **
## Knowledge5. Excellent 0.14418 0.05942 2.427 0.015997 *
## Recent2. Somewhat Confident 0.01959 0.03017 0.649 0.516655
## Recent3. Confident 0.03890 0.03953 0.984 0.326070
## Recent4. Very Confident 0.02933 0.04981 0.589 0.556548
## CHW -0.05927 0.02906 -2.040 0.042497 *
## PrePost2Pre -0.08915 0.02670 -3.340 0.000976 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.147 on 235 degrees of freedom
## Multiple R-squared: 0.4546, Adjusted R-squared: 0.4128
## F-statistic: 10.88 on 18 and 235 DF, p-value: < 2.2e-16hist(mylm$residuals, col="blue", main="Residuals")