Generate the Dataset for Case Crossover Analysis
Description
Generate the dataset for case crossover analysis.
Usage
CXover.data(data,date,ID,direction,apart)Arguments
data
a data.frame containing the date of each case.
date
the name of the variable in the data indicating the date of each case reported to the database.
ID
the name of the variable in the data indicating case ID, if not specified, it will automatically generated starting from 1.
direction
"month4" (default),"pre4" or "after4". With "pre4" (or "after4"), each case day will be matched with same weekdays in previous (or subsequent) 4 weeks. With "month4", each case day will be matched with same weekdays in the same month, which is the most common in literature.
apart
7 (default) or 14. With apart==7, each case day will be 7 days apart from control days in the same month as in the traditional case-crossover design while with apart==14, days will be 14 days apart each other.
Details
Not limited to hospital data, but also applicable to other surveillance data.
Value
dataset
A data.frame ready for the case crossover analysis, with following variables:
ID
same ID represents the same patient.
Date
one case day is matched with 3-4 control days.
status
indicating whether it is a case day or a control day.
References
Zhang W, Lin S, Hopke PK, et al. Triggering of cardiovascular hospital admissions by fine particle concentrations in New York state: Before, during, and after implementation of multiple environmental policies and a recession. Environ. Pollut. 2018;242:1404–1416.
Examples
# similated data
set.seed(2018)
dataset=data.frame(
patient=1:1000,
primdiag=sample(390:398,1000,replace=TRUE),
onset=sample(seq.Date(as.Date("2015/2/1"),as.Date("2016/2/1"),"1 day"),1000,replace=TRUE),
sex=sample(c("M","F"),1000,replace=TRUE),
county=sample(c("Albany","New York"),1000,replace=TRUE))
out.data=CXover.data(data=dataset,date="onset",ID="patient")
head(out.data)
Create a dbf File for Geocoding
Description
Generate address variables and output the data as a dbf file for geocoding in ArcGIS.
Usage
DBFgeocode(data,cityname,roadaddress,mailbox,ZIP)Arguments
data
A data.frame containing address variables that are necessry for geocoding.
cityname
The name of the variable in the data indicating city or county names.
roadaddress
The name of the variable in the data indicating home addresses.
mailbox
Optional address information such as the number of mailbox and the number of floor.
ZIP
The name of the variable in the data indicating ZIP codes.
Value
Users may output the function return to the computer as the dbf file using write.dbf ().
Note
In the dbf file, a variable named "singleline" will be used in the second step of geocoding, while variables roadaddress,cityname and ZIP will be seperately used in the first step, and the variable ZIP for the last step.
Examples
# similated data
datatest=data.frame(county=c("Albany","Albany","Albany"),
address1=c("1 Lincoln ave","2 Lincoln ave","489 Washinton ave"),
address2=c("1st floor","1st floor","2nd floor"),
zip=12206
)
DBFgeocode(data=datatest,cityname="county",roadaddress="address1",
mailbox="address2",ZIP="zip")
Determine the Area that Each Record Is Located in
Description
Identify the residential county/city/census tract for each case, and add county/city/census tract ID.
Usage
FIPS.name(data,ID.case,long.case,lat.case,map,state.map,level.map,areaID)Arguments
data
A data.frame containing the ID and coordinates of cases
ID.case
Name of the variable in the data indicating the case ID.
long.case
Name of the variable in the data indicating the longitude of cases.
lat.case
Name of the variable in the data indicating the latitude of cases.
map
The reference map containing the boundary of county/city/census tract. Do not have to specify for study areas within the U.S. A map for a region outside the U.S. can be imported as a "spatialpolygonsdataframe" object.
state.map
State FIPS code for the study area, e,g, "36" for the New York State. Ignored if readers' own map is being used.
level.map
"county" or "tract", determine whether cases will be macthed to counties or census tracts. Ignored if readers' own map is being used.
areaID
Name of the variable in the map indicating the area ID. Use the default if the study is within the U.S.
Details
Not limited to hospital data, but also applicable to other surveillance data.
Value
areaID
The area unique ID such as FIPS code and ZIP code will be added to the original data.
Examples
set.seed(2018)
dataset=data.frame(Patient=1:2,lat=rnorm(2,42,0.5),long=rnorm(2,-76,1))
data.out=FIPS.name(data=dataset,ID.case="Patient",long.case="long",
lat.case="lat",state.map="36",level.map="tract",areaID="GEOID")
Generate the Case Series
Description
Estimates the daily number of cases reported by multiple grouping factors.
Usage
case.series(data,ICD,diagnosis,date,start,end,by1,by2,by3,by4,by5)
Arguments
data
a data.frame containing with each row representing a case, and each column representing the patient characteristics such as gender, age, admission date, and discharge date, etc.
ICD
a vector of ICD 9, or 10 codes, or a mix of them, which users are willing to calculate the daily numbers for; can be of length 3-6.
diagnosis
the name of the variable in the data containing the diagnostic code upon admission.
date
the name of the variable in the data showing the admission date, either in the format like "20181129" or "2018年11月29日".
start, end
the start and end date for the case series to be generated.
by1, by2, by3, by4, by5
the name of the variable in the data used as grouping variables.
Details
Not limited to hospital data, but also applicable to other surveillance data.
Value
dataset
A case series will be generated for time series analysis, trend analysis and displaying, with following variables:
date
from the start date to the end date as user specified, with 1 day bin.
case
the daily number of cases diagnosed with diseases of user specified ICD codes.
others
grouping variables.
Note
When applied to other medical data without ICD code, users may arbitrarily set a ICD code, meanwhile, define the diagnosis variable in the data to the same ICD code.
Examples
set.seed(2018)
data=data.frame(
patient=1:10000,
primdiag=sample(390:398,10000,replace=TRUE),
onset=sample(seq.Date(as.Date("2015/2/1"),
as.Date("2016/2/1"),"1 day"),10000,replace=TRUE),
sex=sample(c("M","F"),10000,replace=TRUE),
county=sample(c("Albany","New York"),10000,replace=TRUE)
)
output.series=case.series(
data,ICD=392:396,diagnosis="primdiag",
date="onset",start="2015/1/1",end="2016/12/31",by1="sex")
head(output.series)
Generate a Descriptive Table
Description
Generate a comprehensive descriptive table with intergroup comparison.
Usage
desc.comp(data,variables,by,margin,avg.num,test.num)
Arguments
data
a data.frame containing the variables to be described and a group variable
variables
a numeric variable indicating the columns of variables to be described.
by
a number indicating the column of the group variable
margin
calculate the proportion for categorical variables by 1 (row) or 2 (column).
avg.num
"mean", describe continuous variables with mean and standard deviation; "median", describe continuous variables with median and interquantile range; otherwise, normal distribution test will be conducted, for normal distributed variables, "mean" will be used, otherwise, "median" will be used.
test.num
"metric", t test or anova will be used for intergroup comparison; "nonmetric", Wilcoxon rank sum test or Kruskal-Wallis test will be used; otherwise, normal distribution test will be conducted, for normal distributed variables, "metric" will be used, otherwise, "nonmetric" will be used.
Details
Not limited to hospital data, but also applicable to other surveillance data.
Value
A comprehensive descriptive table with statistics and P value for intergroup comparisons.
Examples
desc.comp(CO2,variables=2:5,by=1,margin=1)
Identify Duplicates and Re-admissions
Description
Identify the duplicates and re-admissions in hospital data with subject identifications.
Usage
dupl.readm(data,UniqueID,date,period)Arguments
data
a data.frame containing "UniqueID" and "date"
UniqueID
the name of the variable in the data indicating case ID.
date
the name of the variable in the data indicating the admission/onset date.
period
the time period used to define an re-admission; period=365 by default.
Details
Not limited to hospital data, but also applicable to other surveillance data with "UniqueID" and "date".
Value
id.dupl
indicating whether it is a duplicated record with exactly the same "UniqueID" and "date" as a previous record. In some hospital data,some patients may be reported twice or even more due to insurance issues. For most studies, researchers may remove this kind of duplicates to avoid potential overcounting problems.
onlyone
indicating whether this is the only record with this ID.
Period
the time period between the current visit and the previous one for a patient; 0 for the 1st visit; and NA for those with only one record.
Nadmission
indicating the times of admission, e.g. 1st, 2nd admission; a patient may have more than one 1st admissions if some periods between two visits are greater than e.g. 365 days.
Examples
dataset=data.frame(
ID=c(1,3,4,2,4,6,3,5,7,1),
onset=c("2015/1/1","2016/1/2","2015/5/9",
"2015/12/1","2016/8/2","2015/5/9",
"2015/11/1","2016/3/2","2016/5/9","2015/9/9")
)
out.data=dupl.readm(data=dataset,
UniqueID="ID",date="onset",period=365)
head(out.data)
Calculate Individual and Cumulative Lags for Exposure
Description
Calculate individual and cumulative lag exposure for specific variables. Cumulative lag exposure was calculated by using moving average.
Usage
exposure_lag(data,var,maxlag,ID,Date,lag_suffix)
Arguments
data
A dataframe.
var
Variable names in the dataframe to specify variables to be used for the lag calculation.
maxlag
A number. The max day for calculating the lag exposure.
ID
A variable name. The exposure station ID.
Date
A variable name. A variable indicating the date of exposure measurement.
lag_suffix
A two-length vector indicating the cumulative lag or the individual lag. The first was the suffix for cumulative lag exposure. The second was for individual lag exposure. Default: c('_cu_lag','_si_lag')
Value
It returns a dataframe with calculated individual and cumulative lag exposures. 'var_cu_lag5' means the moving average from lag 0 to lag 5 days. 'var_si_lag5' means the exposure 5 days ago.
References
Deng X, Friedman S, Ryan I, et al. The independent and synergistic impacts of power outages and floods on hospital admissions for multiple diseases [published online ahead of print, 2022 Mar 5]. Sci Total Environ. 2022;828:154305. doi:10.1016/j.scitotenv.2022.154305
Examples
data=data.frame(
ID=rep(1:5,each=5),
Date=seq(as.Date('2022-01-01'),as.Date('2022-01-05'),by='1 day'),
x=rnorm(25)
)
exposure_lag(data,var='x',maxlag=3,ID='ID',Date='Date')
Mediating Analysis
Description
This function provides convenient algorithm to calculate total effect, mediation effect, direct effect and the proportion of mediation effect.
Usage
mediationking(dataset,outcome,mediator,exposure,n.sim)Arguments
dataset
The dataset that is used for analysis.
outcome
The name of the outcome variable in the dataset.
mediator
The name of the mediator in the dataset.
exposure
The name of the exposure factor in the dataset.
n.sim
Times of simulation to estimate 95% confidence intervals.
Details
Please use set.seed() if you want to get a consistent result; this function will be expended to allow more covariates shortly.
Value
Total effect
The total effect of the exposure on the outcome variable.
Indirect effect
The effect of the exposure on the outcome variable that is caused by mediator.
Direct effect
The effect of the exposure on the outcome variable that is caused by factors other than the mediator.
Meditation.proportion
The proportion of the mediation effect.
Examples
set.seed(1)
exposure<-rnorm(20,0,1)
mediator<-rnorm(20,10,1)
outcome<-rnorm(20,10,1)
dataset<-data.frame(outcome,mediator,exposure)
mediationking(dataset,"outcome","mediator","exposure")
Select cases within certain distance around a site
Description
Identify the closest site (e.g. monitoring sites) for each case, and select cases within certain distance around a site, e.g. 15 miles buffer.
Usage
pick.cases(data,long.case,lat.case,long.sites,lat.sites,radius)Arguments
data
a data.frame containing the coordinates of cases.
long.case
the name of variable in the data indicating the longitude of cases.
lat.case
the name of variable in the data indicating the latitude of cases.
long.sites
a numeric vector containing the longitude of sites.
lat.sites
a numeric vector containing the latitude of sites.
radius
radius of the buffer, e.g."15 miles", "20 kms".
Details
Not limited to hospital data, but also applicable to other surveillance data.
Value
which.site
the closest site to the case.
minDIST
the distance of the case to the closest site; in the same unit as "radius".
Select
an indicator of whether a case was within the buffer.
References
Zhang W, Lin S, Hopke PK, et al. Triggering of cardiovascular hospital admissions by fine particle concentrations in New York state: Before, during, and after implementation of multiple environmental policies and a recession. Environ. Pollut. [electronic article]. 2018;242:1404–1416.
Examples
set.seed(2018)
data=data.frame(Patient=1:100,lat=rnorm(100,41,0.5),long=rnorm(100,-76,1))
long.monitor=c(-73.75464,-78.80953,-73.902,-73.82153,-77.54817)
lat.monitor=c(42.64225,42.87691,40.81618,40.73614,43.14618)
data.out=pick.cases(data,long.case="long",lat.case="lat",
long.sites=long.monitor,lat.sites=lat.monitor,radius="30 miles")
data.out
Extract Values from a Raster Map
Description
Crop the raster with the boundary of areas of your interest, and extract the values from the raster to each of these areas.
Usage
raster_extract(rastermap,refmap,ID.var,ID.code,cutpoint)Arguments
rastermap
a raster map containing the information you need, such as the National Land Cover Database 2011.
refmap
"SpatialPolygonsDataFrame" object. A reference map containing the boundary information of your study areas.
ID.var
the name of variable in the refmap indicating the unique ID for each of your study areas.
ID.code
a character vector containing the unique ID for areas that you want to extract the values to. ID.code=ALL" by default where all areas in the reference map are of interest.
cutpoint
a number to dichotomize the values in the raster; specified ONLY when those values are continuous.
Details
Usually for extracting data which are available as rasters such as the land coverage or land usage data.
Value
ID.code
the column indicating the unique ID for each area, followed by the number of cells for each category/colour within that area.
Total cells
the total number of cells within each area.
Examples
library(raster)
set.seed(4715)
rast=raster(matrix(rnorm(500),100,100))
extent(rast)=c(50,100,10,60)
crs(rast)=CRS("+proj=longlat +datum=WGS84")
ref=cbind(x=c(60,80,80,70), y=c(20,25,40,30))
p=Polygon(ref)
ps=Polygons(list(p),ID="ID")
ref=SpatialPolygons(list(ps))
data=data.frame(value=1, ID="10086",row.names="ID")
ref=SpatialPolygonsDataFrame(ref,data)
proj4string(ref)=CRS("+proj=longlat +datum=WGS84")
raster_extract(rastermap=rast,refmap=ref,ID.var="ID",ID.code="ALL",cutpoint=0.5)