The standardized (mean) difference is a measure of distance between two group means in terms of one or more variables. In practice it is often used as a balance measure of individual covariates before and after propensity score matching. As it is standardized, comparison across variables on different scales is possible. For definitions see http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3144483/#s11title .
Standardized mean differences can be easily calculated with tableone. All standardized mean differences in this package are absolute values, thus, there is no directionality.
## tableone package itself
library(tableone)
## PS matching
library(Matching)
## Weighted analysis
library(survey)
## Reorganizing data
library(reshape2)
## plotting
library(ggplot2)
The right heart catheterization dataset is available at http://biostat.mc.vanderbilt.edu/wiki/Main/DataSets . This dataset was originally used in Connors et al. JAMA 1996;276:889-897, and has been made publicly available.
## Right heart cath dataset
rhc <- read.csv("http://biostat.mc.vanderbilt.edu/wiki/pub/Main/DataSets/rhc.csv")
Out of the 50 covariates, 32 have standardized mean differences of greater than 0.1, which is often considered the sign of important covariate imbalance (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3144483/#s11title ).
## Covariates
vars <- c("age","sex","race","edu","income","ninsclas","cat1","das2d3pc","dnr1",
"ca","surv2md1","aps1","scoma1","wtkilo1","temp1","meanbp1","resp1",
"hrt1","pafi1","paco21","ph1","wblc1","hema1","sod1","pot1","crea1",
"bili1","alb1","resp","card","neuro","gastr","renal","meta","hema",
"seps","trauma","ortho","cardiohx","chfhx","dementhx","psychhx",
"chrpulhx","renalhx","liverhx","gibledhx","malighx","immunhx",
"transhx","amihx")
## Construct a table
tabUnmatched <- CreateTableOne(vars = vars, strata = "swang1", data = rhc, test = FALSE)
## Show table with SMD
print(tabUnmatched, smd = TRUE)
Stratified by swang1
No RHC RHC SMD
n 3551 2184
age (mean (sd)) 61.76 (17.29) 60.75 (15.63) 0.061
sex = Male (%) 1914 (53.9) 1278 (58.5) 0.093
race (%) 0.036
black 585 (16.5) 335 (15.3)
other 213 ( 6.0) 142 ( 6.5)
white 2753 (77.5) 1707 (78.2)
edu (mean (sd)) 11.57 (3.13) 11.86 (3.16) 0.091
income (%) 0.142
$11-$25k 713 (20.1) 452 (20.7)
$25-$50k 500 (14.1) 393 (18.0)
> $50k 257 ( 7.2) 194 ( 8.9)
Under $11k 2081 (58.6) 1145 (52.4)
ninsclas (%) 0.194
Medicaid 454 (12.8) 193 ( 8.8)
Medicare 947 (26.7) 511 (23.4)
Medicare & Medicaid 251 ( 7.1) 123 ( 5.6)
No insurance 186 ( 5.2) 136 ( 6.2)
Private 967 (27.2) 731 (33.5)
Private & Medicare 746 (21.0) 490 (22.4)
cat1 (%) 0.583
ARF 1581 (44.5) 909 (41.6)
CHF 247 ( 7.0) 209 ( 9.6)
COPD 399 (11.2) 58 ( 2.7)
Cirrhosis 175 ( 4.9) 49 ( 2.2)
Colon Cancer 6 ( 0.2) 1 ( 0.0)
Coma 341 ( 9.6) 95 ( 4.3)
Lung Cancer 34 ( 1.0) 5 ( 0.2)
MOSF w/Malignancy 241 ( 6.8) 158 ( 7.2)
MOSF w/Sepsis 527 (14.8) 700 (32.1)
das2d3pc (mean (sd)) 20.37 (5.48) 20.70 (5.03) 0.063
dnr1 = Yes (%) 499 (14.1) 155 ( 7.1) 0.228
ca (%) 0.107
Metastatic 261 ( 7.4) 123 ( 5.6)
No 2652 (74.7) 1727 (79.1)
Yes 638 (18.0) 334 (15.3)
surv2md1 (mean (sd)) 0.61 (0.19) 0.57 (0.20) 0.198
aps1 (mean (sd)) 50.93 (18.81) 60.74 (20.27) 0.501
scoma1 (mean (sd)) 22.25 (31.37) 18.97 (28.26) 0.110
wtkilo1 (mean (sd)) 65.04 (29.50) 72.36 (27.73) 0.256
temp1 (mean (sd)) 37.63 (1.74) 37.59 (1.83) 0.021
meanbp1 (mean (sd)) 84.87 (38.87) 68.20 (34.24) 0.455
resp1 (mean (sd)) 28.98 (13.95) 26.65 (14.17) 0.165
hrt1 (mean (sd)) 112.87 (40.94) 118.93 (41.47) 0.147
pafi1 (mean (sd)) 240.63 (116.66) 192.43 (105.54) 0.433
paco21 (mean (sd)) 39.95 (14.24) 36.79 (10.97) 0.249
ph1 (mean (sd)) 7.39 (0.11) 7.38 (0.11) 0.120
wblc1 (mean (sd)) 15.26 (11.41) 16.27 (12.55) 0.084
hema1 (mean (sd)) 32.70 (8.79) 30.51 (7.42) 0.269
sod1 (mean (sd)) 137.04 (7.68) 136.33 (7.60) 0.092
pot1 (mean (sd)) 4.08 (1.04) 4.05 (1.01) 0.027
crea1 (mean (sd)) 1.92 (2.03) 2.47 (2.05) 0.270
bili1 (mean (sd)) 2.00 (4.43) 2.71 (5.33) 0.145
alb1 (mean (sd)) 3.16 (0.67) 2.98 (0.93) 0.230
resp = Yes (%) 1481 (41.7) 632 (28.9) 0.270
card = Yes (%) 1007 (28.4) 924 (42.3) 0.295
neuro = Yes (%) 575 (16.2) 118 ( 5.4) 0.353
gastr = Yes (%) 522 (14.7) 420 (19.2) 0.121
renal = Yes (%) 147 ( 4.1) 148 ( 6.8) 0.116
meta = Yes (%) 172 ( 4.8) 93 ( 4.3) 0.028
hema = Yes (%) 239 ( 6.7) 115 ( 5.3) 0.062
seps = Yes (%) 515 (14.5) 516 (23.6) 0.234
trauma = Yes (%) 18 ( 0.5) 34 ( 1.6) 0.104
ortho = Yes (%) 3 ( 0.1) 4 ( 0.2) 0.027
cardiohx (mean (sd)) 0.16 (0.37) 0.20 (0.40) 0.116
chfhx (mean (sd)) 0.17 (0.37) 0.19 (0.40) 0.069
dementhx (mean (sd)) 0.12 (0.32) 0.07 (0.25) 0.163
psychhx (mean (sd)) 0.08 (0.27) 0.05 (0.21) 0.143
chrpulhx (mean (sd)) 0.22 (0.41) 0.14 (0.35) 0.192
renalhx (mean (sd)) 0.04 (0.20) 0.05 (0.21) 0.032
liverhx (mean (sd)) 0.07 (0.26) 0.06 (0.24) 0.049
gibledhx (mean (sd)) 0.04 (0.19) 0.02 (0.16) 0.070
malighx (mean (sd)) 0.25 (0.43) 0.20 (0.40) 0.101
immunhx (mean (sd)) 0.26 (0.44) 0.29 (0.45) 0.080
transhx (mean (sd)) 0.09 (0.29) 0.15 (0.36) 0.170
amihx (mean (sd)) 0.03 (0.17) 0.04 (0.20) 0.074
## Count covariates with important imbalance
addmargins(table(ExtractSmd(tabUnmatched) > 0.1))
FALSE TRUE Sum
18 32 50
Usually a logistic regression model is used to estimate individual propensity scores. The model here is taken from “How To Use Propensity Score Analysis” (http://www.mc.vanderbilt.edu/crc/workshop_files/2008-04-11.pdf ). Predicted probabilities of being assigned to right heart catherterization, being assigned no right heart catherterization, being assigned to the true assignment, as well as the smaller of the probabilities of being assigned to right heart catherterization or no right heart catherterization are calculated for later use in propensity score matching and weighting.
## Fit model
psModel <- glm(formula = swang1 ~ age + sex + race + edu + income + ninsclas +
cat1 + das2d3pc + dnr1 + ca + surv2md1 + aps1 + scoma1 +
wtkilo1 + temp1 + meanbp1 + resp1 + hrt1 + pafi1 +
paco21 + ph1 + wblc1 + hema1 + sod1 + pot1 + crea1 +
bili1 + alb1 + resp + card + neuro + gastr + renal +
meta + hema + seps + trauma + ortho + cardiohx + chfhx +
dementhx + psychhx + chrpulhx + renalhx + liverhx + gibledhx +
malighx + immunhx + transhx + amihx,
family = binomial(link = "logit"),
data = rhc)
## Predicted probability of being assigned to RHC
rhc$pRhc <- predict(psModel, type = "response")
## Predicted probability of being assigned to no RHC
rhc$pNoRhc <- 1 - rhc$pRhc
## Predicted probability of being assigned to the
## treatment actually assigned (either RHC or no RHC)
rhc$pAssign <- NA
rhc$pAssign[rhc$swang1 == "RHC"] <- rhc$pRhc[rhc$swang1 == "RHC"]
rhc$pAssign[rhc$swang1 == "No RHC"] <- rhc$pNoRhc[rhc$swang1 == "No RHC"]
## Smaller of pRhc vs pNoRhc for matching weight
rhc$pMin <- pmin(rhc$pRhc, rhc$pNoRhc)
The Matching package can be used for propensity score matching. The logit of propensity score is often used as the matching scale, and the matchign caliper is often 0.2 \(\times\) SD(logit(PS)). See http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3144483/#s5title for suggestions. After matching, all the standardized mean differences are below 0.1.
listMatch <- Match(Tr = (rhc$swang1 == "RHC"), # Need to be in 0,1
## logit of PS,i.e., log(PS/(1-PS)) as matching scale
X = log(rhc$pRhc / rhc$pNoRhc),
## 1:1 matching
M = 1,
## caliper = 0.2 * SD(logit(PS))
caliper = 0.2,
replace = FALSE,
ties = TRUE,
version = "fast")
## Extract matched data
rhcMatched <- rhc[unlist(listMatch[c("index.treated","index.control")]), ]
## Construct a table
tabMatched <- CreateTableOne(vars = vars, strata = "swang1", data = rhcMatched, test = FALSE)
## Show table with SMD
print(tabMatched, smd = TRUE)
Stratified by swang1
No RHC RHC SMD
n 1563 1563
age (mean (sd)) 60.92 (17.23) 60.58 (15.72) 0.021
sex = Male (%) 883 (56.5) 892 (57.1) 0.012
race (%) 0.014
black 253 (16.2) 247 (15.8)
other 102 ( 6.5) 99 ( 6.3)
white 1208 (77.3) 1217 (77.9)
edu (mean (sd)) 11.78 (3.16) 11.78 (3.16) 0.001
income (%) 0.037
$11-$25k 317 (20.3) 337 (21.6)
$25-$50k 255 (16.3) 262 (16.8)
> $50k 126 ( 8.1) 124 ( 7.9)
Under $11k 865 (55.3) 840 (53.7)
ninsclas (%) 0.047
Medicaid 171 (10.9) 152 ( 9.7)
Medicare 374 (23.9) 371 (23.7)
Medicare & Medicaid 93 ( 6.0) 94 ( 6.0)
No insurance 80 ( 5.1) 89 ( 5.7)
Private 490 (31.3) 498 (31.9)
Private & Medicare 355 (22.7) 359 (23.0)
cat1 (%) 0.064
ARF 710 (45.4) 679 (43.4)
CHF 167 (10.7) 175 (11.2)
COPD 54 ( 3.5) 57 ( 3.6)
Cirrhosis 46 ( 2.9) 47 ( 3.0)
Colon Cancer 0 ( 0.0) 1 ( 0.1)
Coma 76 ( 4.9) 76 ( 4.9)
Lung Cancer 3 ( 0.2) 5 ( 0.3)
MOSF w/Malignancy 134 ( 8.6) 128 ( 8.2)
MOSF w/Sepsis 373 (23.9) 395 (25.3)
das2d3pc (mean (sd)) 20.38 (5.36) 20.58 (5.08) 0.037
dnr1 = Yes (%) 136 ( 8.7) 130 ( 8.3) 0.014
ca (%) 0.040
Metastatic 112 ( 7.2) 98 ( 6.3)
No 1172 (75.0) 1194 (76.4)
Yes 279 (17.9) 271 (17.3)
surv2md1 (mean (sd)) 0.58 (0.20) 0.59 (0.20) 0.032
aps1 (mean (sd)) 57.11 (19.63) 57.27 (19.66) 0.008
scoma1 (mean (sd)) 18.53 (28.66) 18.85 (28.26) 0.011
wtkilo1 (mean (sd)) 69.89 (26.31) 70.72 (27.19) 0.031
temp1 (mean (sd)) 37.63 (1.88) 37.62 (1.74) 0.003
meanbp1 (mean (sd)) 73.25 (35.48) 73.07 (35.74) 0.005
resp1 (mean (sd)) 28.02 (13.84) 28.05 (14.15) 0.002
hrt1 (mean (sd)) 117.33 (42.45) 117.77 (40.24) 0.011
pafi1 (mean (sd)) 207.87 (106.02) 211.39 (108.01) 0.033
paco21 (mean (sd)) 37.64 (10.89) 37.45 (11.56) 0.017
ph1 (mean (sd)) 7.39 (0.11) 7.39 (0.11) 0.017
wblc1 (mean (sd)) 15.49 (11.69) 15.92 (13.00) 0.035
hema1 (mean (sd)) 30.75 (7.99) 30.91 (7.55) 0.021
sod1 (mean (sd)) 136.55 (7.77) 136.64 (7.43) 0.012
pot1 (mean (sd)) 4.03 (1.03) 4.05 (0.99) 0.015
crea1 (mean (sd)) 2.27 (2.36) 2.28 (1.96) 0.004
bili1 (mean (sd)) 2.56 (5.64) 2.55 (5.09) 0.002
alb1 (mean (sd)) 3.04 (0.70) 3.04 (0.96) 0.004
resp = Yes (%) 525 (33.6) 519 (33.2) 0.008
card = Yes (%) 598 (38.3) 599 (38.3) 0.001
neuro = Yes (%) 109 ( 7.0) 109 ( 7.0) <0.001
gastr = Yes (%) 277 (17.7) 291 (18.6) 0.023
renal = Yes (%) 92 ( 5.9) 94 ( 6.0) 0.005
meta = Yes (%) 68 ( 4.4) 74 ( 4.7) 0.018
hema = Yes (%) 102 ( 6.5) 97 ( 6.2) 0.013
seps = Yes (%) 327 (20.9) 332 (21.2) 0.008
trauma = Yes (%) 15 ( 1.0) 12 ( 0.8) 0.021
ortho = Yes (%) 2 ( 0.1) 1 ( 0.1) 0.021
cardiohx (mean (sd)) 0.20 (0.40) 0.20 (0.40) 0.002
chfhx (mean (sd)) 0.20 (0.40) 0.20 (0.40) 0.014
dementhx (mean (sd)) 0.07 (0.26) 0.07 (0.26) <0.001
psychhx (mean (sd)) 0.05 (0.22) 0.05 (0.23) 0.017
chrpulhx (mean (sd)) 0.15 (0.36) 0.15 (0.36) 0.005
renalhx (mean (sd)) 0.05 (0.23) 0.05 (0.22) 0.020
liverhx (mean (sd)) 0.07 (0.25) 0.07 (0.26) 0.008
gibledhx (mean (sd)) 0.03 (0.17) 0.03 (0.17) 0.011
malighx (mean (sd)) 0.24 (0.43) 0.23 (0.42) 0.036
immunhx (mean (sd)) 0.28 (0.45) 0.28 (0.45) 0.007
transhx (mean (sd)) 0.12 (0.32) 0.12 (0.33) 0.010
amihx (mean (sd)) 0.03 (0.18) 0.03 (0.17) 0.015
## Count covariates with important imbalance
addmargins(table(ExtractSmd(tabMatched) > 0.1))
FALSE Sum
50 50
The matching weight method is a weighting analogue to the 1:1 pairwise algorithmic matching (http://www.ncbi.nlm.nih.gov/pubmed/23902694 ). An earlier version of the paper is available free (http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.359.4724&rep=rep1&type=pdf ). The matching weight is defined as the smaller of the predicted probabilities of receiving or not receiving the treatment over the predicted probability of being assigned to the arm the patient is actually in. After weighting, all the standardized mean differences are below 0.1. The standardized mean differences in weighted data are explained in http://onlinelibrary.wiley.com/doi/10.1002/sim.6607/full .
## Matching weight
rhc$mw <- rhc$pMin / rhc$pAssign
## Weighted data
rhcSvy <- svydesign(ids = ~ 1, data = rhc, weights = ~ mw)
## Construct a table (This is a bit slow.)
tabWeighted <- svyCreateTableOne(vars = vars, strata = "swang1", data = rhcSvy, test = FALSE)
## Show table with SMD
print(tabWeighted, smd = TRUE)
Stratified by swang1
No RHC RHC SMD
n 1522.89 1520.27
age (mean (sd)) 60.82 (17.16) 60.77 (15.79) 0.003
sex = Male (%) 875.8 (57.5) 872.3 (57.4) 0.003
race (%) 0.009
black 238.1 (15.6) 235.8 (15.5)
other 94.9 ( 6.2) 97.8 ( 6.4)
white 1189.9 (78.1) 1186.6 (78.1)
edu (mean (sd)) 11.80 (3.17) 11.80 (3.09) 0.002
income (%) 0.004
$11-$25k 316.5 (20.8) 317.0 (20.9)
$25-$50k 251.7 (16.5) 250.8 (16.5)
> $50k 127.1 ( 8.3) 128.4 ( 8.4)
Under $11k 827.6 (54.3) 824.1 (54.2)
ninsclas (%) 0.014
Medicaid 153.7 (10.1) 151.9 (10.0)
Medicare 361.1 (23.7) 369.0 (24.3)
Medicare & Medicaid 91.5 ( 6.0) 91.2 ( 6.0)
No insurance 85.8 ( 5.6) 86.6 ( 5.7)
Private 487.0 (32.0) 482.2 (31.7)
Private & Medicare 343.7 (22.6) 339.3 (22.3)
cat1 (%) 0.017
ARF 685.8 (45.0) 679.9 (44.7)
CHF 160.1 (10.5) 163.2 (10.7)
COPD 56.2 ( 3.7) 57.2 ( 3.8)
Cirrhosis 45.0 ( 3.0) 47.0 ( 3.1)
Colon Cancer 0.9 ( 0.1) 1.0 ( 0.1)
Coma 79.4 ( 5.2) 77.4 ( 5.1)
Lung Cancer 4.2 ( 0.3) 5.0 ( 0.3)
MOSF w/Malignancy 122.4 ( 8.0) 121.5 ( 8.0)
MOSF w/Sepsis 368.9 (24.2) 368.1 (24.2)
das2d3pc (mean (sd)) 20.58 (5.45) 20.56 (5.05) 0.005
dnr1 = Yes (%) 131.5 ( 8.6) 129.2 ( 8.5) 0.005
ca (%) 0.006
Metastatic 98.6 ( 6.5) 98.0 ( 6.4)
No 1160.5 (76.2) 1162.3 (76.5)
Yes 263.7 (17.3) 259.9 (17.1)
surv2md1 (mean (sd)) 0.58 (0.20) 0.58 (0.20) 0.010
aps1 (mean (sd)) 57.30 (19.53) 57.13 (19.73) 0.008
scoma1 (mean (sd)) 19.12 (29.10) 19.10 (28.51) 0.001
wtkilo1 (mean (sd)) 70.19 (26.54) 70.19 (27.30) <0.001
temp1 (mean (sd)) 37.63 (1.88) 37.64 (1.74) <0.001
meanbp1 (mean (sd)) 73.18 (35.48) 73.22 (35.50) 0.001
resp1 (mean (sd)) 28.16 (13.84) 28.10 (14.09) 0.004
hrt1 (mean (sd)) 116.96 (42.74) 116.71 (40.28) 0.006
pafi1 (mean (sd)) 209.93 (107.48) 210.31 (108.23) 0.004
paco21 (mean (sd)) 37.56 (10.80) 37.51 (11.59) 0.004
ph1 (mean (sd)) 7.39 (0.11) 7.39 (0.11) 0.003
wblc1 (mean (sd)) 15.82 (12.03) 15.69 (12.69) 0.010
hema1 (mean (sd)) 30.90 (8.10) 30.95 (7.57) 0.007
sod1 (mean (sd)) 136.54 (7.86) 136.58 (7.38) 0.005
pot1 (mean (sd)) 4.04 (1.04) 4.05 (0.99) 0.004
crea1 (mean (sd)) 2.27 (2.31) 2.27 (1.95) <0.001
bili1 (mean (sd)) 2.50 (5.37) 2.54 (5.15) 0.008
alb1 (mean (sd)) 3.04 (0.70) 3.04 (0.97) <0.001
resp = Yes (%) 516.6 (33.9) 512.6 (33.7) 0.004
card = Yes (%) 582.2 (38.2) 585.6 (38.5) 0.006
neuro = Yes (%) 109.6 ( 7.2) 109.0 ( 7.2) 0.001
gastr = Yes (%) 270.3 (17.8) 272.7 (17.9) 0.005
renal = Yes (%) 89.5 ( 5.9) 90.7 ( 6.0) 0.004
meta = Yes (%) 70.0 ( 4.6) 70.2 ( 4.6) 0.001
hema = Yes (%) 93.5 ( 6.1) 95.0 ( 6.2) 0.004
seps = Yes (%) 325.5 (21.4) 322.0 (21.2) 0.005
trauma = Yes (%) 14.8 ( 1.0) 14.3 ( 0.9) 0.003
ortho = Yes (%) 1.0 ( 0.1) 0.9 ( 0.1) 0.003
cardiohx (mean (sd)) 0.20 (0.40) 0.20 (0.40) <0.001
chfhx (mean (sd)) 0.20 (0.40) 0.20 (0.40) 0.004
dementhx (mean (sd)) 0.08 (0.26) 0.08 (0.26) 0.003
psychhx (mean (sd)) 0.05 (0.23) 0.05 (0.22) 0.004
chrpulhx (mean (sd)) 0.16 (0.36) 0.16 (0.36) 0.001
renalhx (mean (sd)) 0.05 (0.22) 0.05 (0.22) 0.001
liverhx (mean (sd)) 0.07 (0.25) 0.07 (0.25) 0.003
gibledhx (mean (sd)) 0.03 (0.17) 0.03 (0.17) 0.007
malighx (mean (sd)) 0.23 (0.42) 0.23 (0.42) 0.007
immunhx (mean (sd)) 0.28 (0.45) 0.28 (0.45) <0.001
transhx (mean (sd)) 0.12 (0.33) 0.12 (0.33) 0.004
amihx (mean (sd)) 0.03 (0.18) 0.03 (0.18) 0.006
## Count covariates with important imbalance
addmargins(table(ExtractSmd(tabWeighted) > 0.1))
FALSE Sum
50 50
A plot showing covariate balance is often constructed to demonstrate the balancing effect of matching and/or weighting. Given the same propensity score model, the matching weight method often achieves better covariate balance than matching.
## Construct a data frame containing variable name and SMD from all methods
dataPlot <- data.frame(variable = names(ExtractSmd(tabUnmatched)),
Unmatched = ExtractSmd(tabUnmatched),
Matched = ExtractSmd(tabMatched),
Weighted = ExtractSmd(tabWeighted))
## Create long-format data for ggplot2
dataPlotMelt <- melt(data = dataPlot,
id.vars = c("variable"),
variable.name = "Method",
value.name = "SMD")
## Order variable names by magnitude of SMD
varNames <- as.character(dataPlot$variable)[order(dataPlot$Unmatched)]
## Order factor levels in the same order
dataPlotMelt$variable <- factor(dataPlotMelt$variable,
levels = varNames)
## Plot using ggplot2
ggplot(data = dataPlotMelt, mapping = aes(x = variable, y = SMD,
group = Method, color = Method)) +
layer(geom = "line") +
layer(geom = "point") +
layer(geom = "hline", yintercept = 0.1, color = "black", size = 0.1) +
coord_flip() +
theme_bw() + theme(legend.key = element_blank())
To construct a side-by-side table, data can be extracted as a matrix and combined using the print() method, which actually invisibly returns a matrix.
## Column bind tables
resCombo <- cbind(print(tabUnmatched, printToggle = FALSE),
print(tabMatched, printToggle = FALSE),
print(tabWeighted, printToggle = FALSE))
## Add group name row, and rewrite column names
resCombo <- rbind(Group = rep(c("No RHC","RHC"), 3), resCombo)
colnames(resCombo) <- c("Unmatched","","Matched","","Weighted","")
print(resCombo, quote = FALSE)
Unmatched Matched Weighted
Group No RHC RHC No RHC RHC No RHC RHC
n 3551 2184 1563 1563 1522.89 1520.27
age (mean (sd)) 61.76 (17.29) 60.75 (15.63) 60.92 (17.23) 60.58 (15.72) 60.82 (17.16) 60.77 (15.79)
sex = Male (%) 1914 (53.9) 1278 (58.5) 883 (56.5) 892 (57.1) 875.8 (57.5) 872.3 (57.4)
race (%)
black 585 (16.5) 335 (15.3) 253 (16.2) 247 (15.8) 238.1 (15.6) 235.8 (15.5)
other 213 ( 6.0) 142 ( 6.5) 102 ( 6.5) 99 ( 6.3) 94.9 ( 6.2) 97.8 ( 6.4)
white 2753 (77.5) 1707 (78.2) 1208 (77.3) 1217 (77.9) 1189.9 (78.1) 1186.6 (78.1)
edu (mean (sd)) 11.57 (3.13) 11.86 (3.16) 11.78 (3.16) 11.78 (3.16) 11.80 (3.17) 11.80 (3.09)
income (%)
$11-$25k 713 (20.1) 452 (20.7) 317 (20.3) 337 (21.6) 316.5 (20.8) 317.0 (20.9)
$25-$50k 500 (14.1) 393 (18.0) 255 (16.3) 262 (16.8) 251.7 (16.5) 250.8 (16.5)
> $50k 257 ( 7.2) 194 ( 8.9) 126 ( 8.1) 124 ( 7.9) 127.1 ( 8.3) 128.4 ( 8.4)
Under $11k 2081 (58.6) 1145 (52.4) 865 (55.3) 840 (53.7) 827.6 (54.3) 824.1 (54.2)
ninsclas (%)
Medicaid 454 (12.8) 193 ( 8.8) 171 (10.9) 152 ( 9.7) 153.7 (10.1) 151.9 (10.0)
Medicare 947 (26.7) 511 (23.4) 374 (23.9) 371 (23.7) 361.1 (23.7) 369.0 (24.3)
Medicare & Medicaid 251 ( 7.1) 123 ( 5.6) 93 ( 6.0) 94 ( 6.0) 91.5 ( 6.0) 91.2 ( 6.0)
No insurance 186 ( 5.2) 136 ( 6.2) 80 ( 5.1) 89 ( 5.7) 85.8 ( 5.6) 86.6 ( 5.7)
Private 967 (27.2) 731 (33.5) 490 (31.3) 498 (31.9) 487.0 (32.0) 482.2 (31.7)
Private & Medicare 746 (21.0) 490 (22.4) 355 (22.7) 359 (23.0) 343.7 (22.6) 339.3 (22.3)
cat1 (%)
ARF 1581 (44.5) 909 (41.6) 710 (45.4) 679 (43.4) 685.8 (45.0) 679.9 (44.7)
CHF 247 ( 7.0) 209 ( 9.6) 167 (10.7) 175 (11.2) 160.1 (10.5) 163.2 (10.7)
COPD 399 (11.2) 58 ( 2.7) 54 ( 3.5) 57 ( 3.6) 56.2 ( 3.7) 57.2 ( 3.8)
Cirrhosis 175 ( 4.9) 49 ( 2.2) 46 ( 2.9) 47 ( 3.0) 45.0 ( 3.0) 47.0 ( 3.1)
Colon Cancer 6 ( 0.2) 1 ( 0.0) 0 ( 0.0) 1 ( 0.1) 0.9 ( 0.1) 1.0 ( 0.1)
Coma 341 ( 9.6) 95 ( 4.3) 76 ( 4.9) 76 ( 4.9) 79.4 ( 5.2) 77.4 ( 5.1)
Lung Cancer 34 ( 1.0) 5 ( 0.2) 3 ( 0.2) 5 ( 0.3) 4.2 ( 0.3) 5.0 ( 0.3)
MOSF w/Malignancy 241 ( 6.8) 158 ( 7.2) 134 ( 8.6) 128 ( 8.2) 122.4 ( 8.0) 121.5 ( 8.0)
MOSF w/Sepsis 527 (14.8) 700 (32.1) 373 (23.9) 395 (25.3) 368.9 (24.2) 368.1 (24.2)
das2d3pc (mean (sd)) 20.37 (5.48) 20.70 (5.03) 20.38 (5.36) 20.58 (5.08) 20.58 (5.45) 20.56 (5.05)
dnr1 = Yes (%) 499 (14.1) 155 ( 7.1) 136 ( 8.7) 130 ( 8.3) 131.5 ( 8.6) 129.2 ( 8.5)
ca (%)
Metastatic 261 ( 7.4) 123 ( 5.6) 112 ( 7.2) 98 ( 6.3) 98.6 ( 6.5) 98.0 ( 6.4)
No 2652 (74.7) 1727 (79.1) 1172 (75.0) 1194 (76.4) 1160.5 (76.2) 1162.3 (76.5)
Yes 638 (18.0) 334 (15.3) 279 (17.9) 271 (17.3) 263.7 (17.3) 259.9 (17.1)
surv2md1 (mean (sd)) 0.61 (0.19) 0.57 (0.20) 0.58 (0.20) 0.59 (0.20) 0.58 (0.20) 0.58 (0.20)
aps1 (mean (sd)) 50.93 (18.81) 60.74 (20.27) 57.11 (19.63) 57.27 (19.66) 57.30 (19.53) 57.13 (19.73)
scoma1 (mean (sd)) 22.25 (31.37) 18.97 (28.26) 18.53 (28.66) 18.85 (28.26) 19.12 (29.10) 19.10 (28.51)
wtkilo1 (mean (sd)) 65.04 (29.50) 72.36 (27.73) 69.89 (26.31) 70.72 (27.19) 70.19 (26.54) 70.19 (27.30)
temp1 (mean (sd)) 37.63 (1.74) 37.59 (1.83) 37.63 (1.88) 37.62 (1.74) 37.63 (1.88) 37.64 (1.74)
meanbp1 (mean (sd)) 84.87 (38.87) 68.20 (34.24) 73.25 (35.48) 73.07 (35.74) 73.18 (35.48) 73.22 (35.50)
resp1 (mean (sd)) 28.98 (13.95) 26.65 (14.17) 28.02 (13.84) 28.05 (14.15) 28.16 (13.84) 28.10 (14.09)
hrt1 (mean (sd)) 112.87 (40.94) 118.93 (41.47) 117.33 (42.45) 117.77 (40.24) 116.96 (42.74) 116.71 (40.28)
pafi1 (mean (sd)) 240.63 (116.66) 192.43 (105.54) 207.87 (106.02) 211.39 (108.01) 209.93 (107.48) 210.31 (108.23)
paco21 (mean (sd)) 39.95 (14.24) 36.79 (10.97) 37.64 (10.89) 37.45 (11.56) 37.56 (10.80) 37.51 (11.59)
ph1 (mean (sd)) 7.39 (0.11) 7.38 (0.11) 7.39 (0.11) 7.39 (0.11) 7.39 (0.11) 7.39 (0.11)
wblc1 (mean (sd)) 15.26 (11.41) 16.27 (12.55) 15.49 (11.69) 15.92 (13.00) 15.82 (12.03) 15.69 (12.69)
hema1 (mean (sd)) 32.70 (8.79) 30.51 (7.42) 30.75 (7.99) 30.91 (7.55) 30.90 (8.10) 30.95 (7.57)
sod1 (mean (sd)) 137.04 (7.68) 136.33 (7.60) 136.55 (7.77) 136.64 (7.43) 136.54 (7.86) 136.58 (7.38)
pot1 (mean (sd)) 4.08 (1.04) 4.05 (1.01) 4.03 (1.03) 4.05 (0.99) 4.04 (1.04) 4.05 (0.99)
crea1 (mean (sd)) 1.92 (2.03) 2.47 (2.05) 2.27 (2.36) 2.28 (1.96) 2.27 (2.31) 2.27 (1.95)
bili1 (mean (sd)) 2.00 (4.43) 2.71 (5.33) 2.56 (5.64) 2.55 (5.09) 2.50 (5.37) 2.54 (5.15)
alb1 (mean (sd)) 3.16 (0.67) 2.98 (0.93) 3.04 (0.70) 3.04 (0.96) 3.04 (0.70) 3.04 (0.97)
resp = Yes (%) 1481 (41.7) 632 (28.9) 525 (33.6) 519 (33.2) 516.6 (33.9) 512.6 (33.7)
card = Yes (%) 1007 (28.4) 924 (42.3) 598 (38.3) 599 (38.3) 582.2 (38.2) 585.6 (38.5)
neuro = Yes (%) 575 (16.2) 118 ( 5.4) 109 ( 7.0) 109 ( 7.0) 109.6 ( 7.2) 109.0 ( 7.2)
gastr = Yes (%) 522 (14.7) 420 (19.2) 277 (17.7) 291 (18.6) 270.3 (17.8) 272.7 (17.9)
renal = Yes (%) 147 ( 4.1) 148 ( 6.8) 92 ( 5.9) 94 ( 6.0) 89.5 ( 5.9) 90.7 ( 6.0)
meta = Yes (%) 172 ( 4.8) 93 ( 4.3) 68 ( 4.4) 74 ( 4.7) 70.0 ( 4.6) 70.2 ( 4.6)
hema = Yes (%) 239 ( 6.7) 115 ( 5.3) 102 ( 6.5) 97 ( 6.2) 93.5 ( 6.1) 95.0 ( 6.2)
seps = Yes (%) 515 (14.5) 516 (23.6) 327 (20.9) 332 (21.2) 325.5 (21.4) 322.0 (21.2)
trauma = Yes (%) 18 ( 0.5) 34 ( 1.6) 15 ( 1.0) 12 ( 0.8) 14.8 ( 1.0) 14.3 ( 0.9)
ortho = Yes (%) 3 ( 0.1) 4 ( 0.2) 2 ( 0.1) 1 ( 0.1) 1.0 ( 0.1) 0.9 ( 0.1)
cardiohx (mean (sd)) 0.16 (0.37) 0.20 (0.40) 0.20 (0.40) 0.20 (0.40) 0.20 (0.40) 0.20 (0.40)
chfhx (mean (sd)) 0.17 (0.37) 0.19 (0.40) 0.20 (0.40) 0.20 (0.40) 0.20 (0.40) 0.20 (0.40)
dementhx (mean (sd)) 0.12 (0.32) 0.07 (0.25) 0.07 (0.26) 0.07 (0.26) 0.08 (0.26) 0.08 (0.26)
psychhx (mean (sd)) 0.08 (0.27) 0.05 (0.21) 0.05 (0.22) 0.05 (0.23) 0.05 (0.23) 0.05 (0.22)
chrpulhx (mean (sd)) 0.22 (0.41) 0.14 (0.35) 0.15 (0.36) 0.15 (0.36) 0.16 (0.36) 0.16 (0.36)
renalhx (mean (sd)) 0.04 (0.20) 0.05 (0.21) 0.05 (0.23) 0.05 (0.22) 0.05 (0.22) 0.05 (0.22)
liverhx (mean (sd)) 0.07 (0.26) 0.06 (0.24) 0.07 (0.25) 0.07 (0.26) 0.07 (0.25) 0.07 (0.25)
gibledhx (mean (sd)) 0.04 (0.19) 0.02 (0.16) 0.03 (0.17) 0.03 (0.17) 0.03 (0.17) 0.03 (0.17)
malighx (mean (sd)) 0.25 (0.43) 0.20 (0.40) 0.24 (0.43) 0.23 (0.42) 0.23 (0.42) 0.23 (0.42)
immunhx (mean (sd)) 0.26 (0.44) 0.29 (0.45) 0.28 (0.45) 0.28 (0.45) 0.28 (0.45) 0.28 (0.45)
transhx (mean (sd)) 0.09 (0.29) 0.15 (0.36) 0.12 (0.32) 0.12 (0.33) 0.12 (0.33) 0.12 (0.33)
amihx (mean (sd)) 0.03 (0.17) 0.04 (0.20) 0.03 (0.18) 0.03 (0.17) 0.03 (0.18) 0.03 (0.18)
The final analysis can be conducted using matched and weighted data. The results from the matching and matching weight are similar. ShowRegTable() function may come in handly.
## Unmatched model (unadjsuted)
glmUnmatched <- glm(formula = (death == "Yes") ~ swang1,
family = binomial(link = "logit"),
data = rhc)
## Matched model
glmMatched <- glm(formula = (death == "Yes") ~ swang1,
family = binomial(link = "logit"),
data = rhcMatched)
## Weighted model
glmWeighted <- svyglm(formula = (death == "Yes") ~ swang1,
family = binomial(link = "logit"),
design = rhcSvy)
## Show results together
resTogether <- list(Unmatched = ShowRegTable(glmUnmatched, printToggle = FALSE),
Matched = ShowRegTable(glmMatched, printToggle = FALSE),
Weighted = ShowRegTable(glmWeighted, printToggle = FALSE))
print(resTogether, quote = FALSE)
$Unmatched
exp(coef) [confint] p
(Intercept) 1.70 [1.59, 1.82] <0.001
swang1RHC 1.25 [1.12, 1.40] <0.001
$Matched
exp(coef) [confint] p
(Intercept) 1.73 [1.56, 1.92] <0.001
swang1RHC 1.30 [1.12, 1.51] 0.001
$Weighted
exp(coef) [confint] p
(Intercept) 1.70 [1.56, 1.85] <0.001
swang1RHC 1.31 [1.14, 1.49] <0.001