MH readmits - EDA
Nayef Ahmad
February 5, 2019
Read in cleaned data (39179 rows)
df1.input.data <- read_csv(here("results",
"code output",
"2018-12-18_mh-readmissions-dataset-for-classification.csv"))
library(lubridate)
# some more cleaning:
df2.cleaned.data <-
df1.input.data %>%
select(denom_PHN,
Gender,
AdmissionAge,
LHAName,
PostalCode,
GPCode,
AdjustedAdmissionDate,
denom_Institution,
denom_AdmissionDate,
denom_DischargeDate,
AdmissionFacilityLongName,
denom_nursingUnitCode,
AdmissionNursingUnit,
AdmittingDrService,
AdmissionAttendingDoctorService,
AdmissionCategoryCode,
AdmissionCategoryGroup,
AdmissionPatientServiceDADDescription,
ED_visits_all_time:ED_admits_last_365days,
elective_admits_all_time:elective_admits_last_365_days,
num_schizophrenia_dx1_all_time,
Readmit_Flag) %>%
mutate(LOS = difftime(denom_DischargeDate,
denom_AdmissionDate,
units = "days") %>%
as.numeric(),
AdmissionAge = as.numeric(AdmissionAge),
AdjustedAdmissionDate = ymd(AdjustedAdmissionDate)
) %>%
# convert all chars to factors:
mutate_if(is.character, factor) %>%
rename(Institution = denom_Institution,
AdmitAge = AdmissionAge,
ReadmitFlag = Readmit_Flag) %>%
clean_names()
# view results:
str(df2.cleaned.data) # 39179 obs. of 33 variables
head(df2.cleaned.data)
summary(df2.cleaned.data)
# skim(df2.cleaned.data)1. Missing values in field adjusted_admission_date - 786 cases
# df2.cleaned.data %>%
# filter(is.na(adjusted_admission_date)) %>% View("missing vals")
df2.cleaned.data %>%
filter(is.na(adjusted_admission_date)) %>%
select(admit_age) %>%
summary # 786 cases
df2.cleaned.data %>%
filter(admission_facility_long_name == "NULL") %>%
select(admission_facility_long_name) %>%
summary # 800 cases
df2.cleaned.data %>%
filter(gender == "NULL") %>%
select(gender) %>%
summary # 786 casestodo: delete these cases
- These 786 cases should be removed. Don’t really know what’s going on, but there’s important info missing. e.g. all have Gender = NULL. It’s possible that these are weird PatientIDs that aren’t joining properly across data sources.
- Then again: perhaps this is useful information. There may be patterns specific to these patients who have weird non-matching PatientIDs??
Recall that in SQL script 02_mh-readmissions_adding-predictor-fields.sql, all joins across data sources (ED, ADTC, ADR, Community) are done using PatientID, not PHN.
2. Num cases by site, with median and 90th percentile
df2.cleaned.data %>%
group_by(institution) %>%
summarise(median = median(los),
perc.90 = quantile(los, 0.90),
num_cases = n()) %>%
arrange(desc(num_cases)) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped")) | institution | median | perc.90 | num_cases |
|---|---|---|---|
| VGH | 7 | 35.0 | 12401 |
| SPH | 7 | 40.0 | 12010 |
| RHS | 2 | 27.0 | 4987 |
| LGH | 6 | 32.0 | 4971 |
| UBCH | 24 | 60.0 | 1872 |
| SMH | 4 | 20.0 | 1552 |
| PRGH | 7 | 32.9 | 782 |
| BCGH | 2 | 6.0 | 247 |
| SGH | 2 | 8.0 | 243 |
| RWLMH | 2 | 3.7 | 114 |
3. LOS = 0 cases (1886 rows)
# summary(df2.cleaned.data$los)
# ecdf(df2.cleaned.data %>% pull(los)) %>% plot
df3.los.zero <-
df2.cleaned.data %>%
filter(los == 0) %>%
select(institution,
denom_admission_date,
denom_discharge_date,
admit_age,
denom_nursing_unit_code,
admission_nursing_unit,
admitting_dr_service) # %>% nrow() # 1886 rows
df3.los.zero$denom_nursing_unit_code %>% summary
df3.los.zero %>%
select(institution,
denom_nursing_unit_code,
denom_admission_date,
denom_discharge_date) %>%
sample_n(20)df2.cleaned.data %>%
select(los,
institution) %>%
filter(los <= 30) %>%
ggplot(aes(x = los)) +
geom_density() +
facet_wrap(~institution)
df3.los.zero %>%
group_by(denom_nursing_unit_code) %>%
summarize(count = n()) %>%
arrange(desc(count)) %>%
kable() %>%
kable_styling() %>%
scroll_box(width = "100%", height = "400px")| denom_nursing_unit_code | count |
|---|---|
| EMRG | 480 |
| REMR | 365 |
| EIP | 227 |
| EMERB | 196 |
| EMERA | 130 |
| EMERU | 116 |
| EA-SM | 85 |
| EMERC | 63 |
| EMERR | 48 |
| NULL | 39 |
| PER/INPT | 16 |
| UD1E | 15 |
| PAU | 11 |
| PPSY | 9 |
| MH-SM | 8 |
| MERS | 7 |
| EMERD | 6 |
| UD1W | 6 |
| WP3 | 6 |
| 4NW | 5 |
| EMERT | 4 |
| HLD | 4 |
| RPEU | 4 |
| SAU | 3 |
| 2A-SM | 2 |
| 2E | 2 |
| EMERM | 2 |
| ICU | 2 |
| M1S | 2 |
| MIU | 2 |
| SG1 | 2 |
| UD2S | 2 |
| WP2 | 2 |
| 1AA-SM | 1 |
| 3E | 1 |
| A2 | 1 |
| HBU | 1 |
| HCE1 | 1 |
| HCE2 | 1 |
| HCW1 | 1 |
| M3C | 1 |
| N/A | 1 |
| PAR | 1 |
| PED | 1 |
| R2W | 1 |
| RAMB | 1 |
| S5MH | 1 |
| USPO | 1 |
Proportion by site:
df2.cleaned.data %>%
group_by(institution) %>%
summarise(num.zero.los = sum(ifelse(los == 0, 1,0)),
other.los = sum(ifelse(los > 0, 1, 0))) %>%
mutate(prop = num.zero.los/(num.zero.los + other.los),
total = num.zero.los + other.los) %>%
arrange(desc(total)) %>%
kable() %>%
kable_styling()| institution | num.zero.los | other.los | prop | total |
|---|---|---|---|---|
| VGH | 525 | 11876 | 0.0423353 | 12401 |
| SPH | 590 | 11420 | 0.0491257 | 12010 |
| RHS | 376 | 4611 | 0.0753960 | 4987 |
| LGH | 239 | 4732 | 0.0480789 | 4971 |
| UBCH | 8 | 1864 | 0.0042735 | 1872 |
| SMH | 102 | 1450 | 0.0657216 | 1552 |
| PRGH | 25 | 757 | 0.0319693 | 782 |
| BCGH | 5 | 242 | 0.0202429 | 247 |
| SGH | 11 | 232 | 0.0452675 | 243 |
| RWLMH | 5 | 109 | 0.0438596 | 114 |
 Â
Tentative conclusion: the LOS = 0 cases are mostly patients who were ED inpatients. That is, a decision to admit was made, but they were discharged before they ever moved up into the nursing units.
todo: remove LOS = 0 cases?
We may want to exclude these patients in most of the rest of the analysis.
3.1 Create nested dataframe by institution
# nest by site:
df4.site.nested <-
df2.cleaned.data %>%
group_by(institution) %>%
nest()
# add column to find overall readmit rate:
# load function:
source(here::here("code",
"readmit-rates-overall_function.R"))
df4.site.nested %<>%
mutate(readmission_overall = map(data,
readmit_rates_overall_function,
exclude_los_zero = 1,
last_2_years = 1))
# result:
# df4.site.nested$readmission_overall
# df4.site.nested %>% unnest(readmission_overall)
# extract overall readmit rate:
df4.site.nested %<>%
mutate(overall_readmit = map_dbl(readmission_overall,
function(x){x %>% pull(prop)})) %>%
select(institution,
overall_readmit,
everything())4. Distributions with and without LOS = 0 cases
# function for plotting densities:
source(here::here("code",
"plot-density_function.R"))
# function for plotting histograms:
source(here::here("code",
"plot-histogram_function.R"))
# map function across nested df:
df4.site.nested %<>%
mutate(densities = map2(data,
institution,
~plot_density_function(..1,
..2,
exclude_los_zero = 1)),
# now add histograms:
histograms = map2(data,
institution,
~plot_histogram_function(..1,
..2,
exclude_los_zero = 0)))Densities
# view densities:
df4.site.nested$densities## [[1]]
##
## [[2]]
##
## [[3]]
##
## [[4]]
##
## [[5]]
##
## [[6]]
##
## [[7]]
##
## [[8]]
##
## [[9]]
##
## [[10]]
Histograms
df4.site.nested$histograms## [[1]]
##
## [[2]]
##
## [[3]]
##
## [[4]]
##
## [[5]]
##
## [[6]]
##
## [[7]]
##
## [[8]]
##
## [[9]]
##
## [[10]]
# densities:
# pdf(here::here("results",
# "code output",
# "2019-02-12_los-densities_since-2016.pdf"),
# width = 10)
# df4.site.nested$densities
# dev.off()
# histograms:
# pdf(here::here("results",
# "code output",
# "2019-02-12_los-histograms_since-2016.pdf"),
# width = 10)
# df4.site.nested$histograms
# dev.off()5. Age categories
Are there natural splits in age suggested by the data?
# function for plotting age cdfs:
source(here::here("code",
"plot-age-cdf_function.R"))
# add plots in a column:
df4.site.nested %<>%
mutate(age_ecdf = map2(data,
institution,
plot_age_cdf_function))# results:
df4.site.nested$age_ecdf## [[1]]
##
## [[2]]
##
## [[3]]
##
## [[4]]
##
## [[5]]
##
## [[6]]
##
## [[7]]
##
## [[8]]
##
## [[9]]
##
## [[10]]
# save outputs:
# pdf(here::here("results",
# "code output",
# "2019-02-12_age-ecdf.pdf"))
# df4.site.nested$age_ecdf
# dev.off()Segmenting LOS by age quantiles
# function for boxplot by quartile:
source(here::here("code",
"los-boxplot_function.R"))
# map function across nesed df:
df4.site.nested %<>%
mutate(los_boxplot = map2(data,
institution,
los_boxplot_function,
exclude_los_zero = 1))
# print boxplots:
# df4.site.nested$los_boxplot[[1]]
df4.site.nested$los_boxplot## [[1]]
##
## [[2]]
##
## [[3]]
##
## [[4]]
##
## [[5]]
##
## [[6]]
##
## [[7]]
##
## [[8]]
##
## [[9]]
##
## [[10]]
# save outputs:
# pdf(here::here("results",
# "code output",
# "2019-02-13_los-distribution-by-age-quartile.pdf"))
# df4.site.nested$los_boxplot
# dev.off()6. Problems with BCGH and RWLMH
- Note that age is missing in all cases, as well as gender
- Explanation: PatientID in ADR isn’t matching with the PatientIDs in ADTC for these sites?
df2.cleaned.data %>%
filter(institution == "BCGH") %>%
summary() ## denom_phn gender admit_age lha_name
## 9026468137:52 F : 0 Min. : NA NULL :247
## 9023956948:38 M : 0 1st Qu.: NA 100 Mile House : 0
## 9020625092:23 NULL:247 Median : NA Abbotsford : 0
## 9014980462:17 Mean :NaN Agassiz-Harrison : 0
## 9020458731:11 3rd Qu.: NA Alberni : 0
## 9064682565:10 Max. : NA Armstrong-Spallumcheen: 0
## (Other) :96 NA's :247 (Other) : 0
## postal_code gp_code adjusted_admission_date institution
## NULL :247 NULL :170 Min. :NA BCGH :247
## A0A0A0 : 0 23440 : 7 1st Qu.:NA LGH : 0
## OutofC : 0 63606 : 6 Median :NA PRGH : 0
## OutofP : 0 63195 : 3 Mean :NA RHS : 0
## UnkBC : 0 4559 : 1 3rd Qu.:NA RWLMH : 0
## V0A1H0 : 0 (Other): 3 Max. :NA SGH : 0
## (Other): 0 NA's : 57 NA's :247 (Other): 0
## denom_admission_date denom_discharge_date
## Min. :2012-01-10 Min. :2012-04-08
## 1st Qu.:2013-12-19 1st Qu.:2013-12-21
## Median :2015-03-27 Median :2015-03-31
## Mean :2015-04-10 Mean :2015-04-14
## 3rd Qu.:2016-09-18 3rd Qu.:2016-09-20
## Max. :2018-02-16 Max. :2018-02-17
##
## admission_facility_long_name denom_nursing_unit_code
## NULL :247 NULL :247
## Holy Family Hospital : 0 10A : 0
## Lions Gate Hospital : 0 10C : 0
## Mount Saint Joseph Hospital : 0 1AA-SM : 0
## Powell River General Hospital: 0 2A-SM : 0
## Richmond Hospital : 0 2E : 0
## (Other) : 0 (Other): 0
## admission_nursing_unit
## NULL :247
## 1AA-SM : 0
## 2 East - Med/ Post Coronary Care: 0
## 2A-SM : 0
## 3 East - Paediatrics : 0
## 3 West - Maternity : 0
## (Other) : 0
## admitting_dr_service admission_attending_doctor_service
## NULL :247 NULL :247
## Anaesthesiology : 0 Anaesthesiology : 0
## Anatomical Pathology : 0 Cardiac Surgery : 0
## Cardiac Surgery : 0 Cardiology : 0
## Cardiology : 0 Critical Care Medicine: 0
## Critical Care Medicine: 0 Diagnostic Radiology : 0
## (Other) : 0 (Other) : 0
## admission_category_code admission_category_group
## NULL :247 Elective : 0
## 1 : 0 Emergent : 0
## 18984780: 0 Newborn : 0
## 2 : 0 NULL :247
## 3 : 0 Urgent : 0
## 309203 : 0 Urgent\\Emergent: 0
## (Other) : 0
## admission_patient_service_dad_description ed_visits_all_time
## NULL :247 Min. : 0.0000
## Alternate Level of Care: 0 1st Qu.: 0.0000
## Cardiology : 0 Median : 0.0000
## Cardiovascular Surgery : 0 Mean : 0.8057
## Family Practice : 0 3rd Qu.: 1.0000
## Gastro-Enterology : 0 Max. :20.0000
## (Other) : 0
## ed_visits_last_30days ed_visits_last_180days ed_visits_last_365days
## Min. :0.00000 Min. : 0.0000 Min. : 0.0000
## 1st Qu.:0.00000 1st Qu.: 0.0000 1st Qu.: 0.0000
## Median :0.00000 Median : 0.0000 Median : 0.0000
## Mean :0.01619 Mean : 0.1457 Mean : 0.2186
## 3rd Qu.:0.00000 3rd Qu.: 0.0000 3rd Qu.: 0.0000
## Max. :2.00000 Max. :13.0000 Max. :14.0000
##
## ed_admits_all_time ed_admits_last_30days ed_admits_last_180days
## Min. :0.0000 Min. :0.000000 Min. :0.00000
## 1st Qu.:0.0000 1st Qu.:0.000000 1st Qu.:0.00000
## Median :0.0000 Median :0.000000 Median :0.00000
## Mean :0.2024 Mean :0.004049 Mean :0.04453
## 3rd Qu.:0.0000 3rd Qu.:0.000000 3rd Qu.:0.00000
## Max. :3.0000 Max. :1.000000 Max. :1.00000
##
## ed_admits_last_365days elective_admits_all_time
## Min. :0.00000 Min. :0.0000
## 1st Qu.:0.00000 1st Qu.:0.0000
## Median :0.00000 Median :0.0000
## Mean :0.06883 Mean :0.3846
## 3rd Qu.:0.00000 3rd Qu.:0.0000
## Max. :1.00000 Max. :6.0000
##
## elective_admits_last_30_days elective_admits_last_180_days
## Min. :0.00000 Min. :0.00000
## 1st Qu.:0.00000 1st Qu.:0.00000
## Median :0.00000 Median :0.00000
## Mean :0.01215 Mean :0.06073
## 3rd Qu.:0.00000 3rd Qu.:0.00000
## Max. :1.00000 Max. :1.00000
##
## elective_admits_last_365_days num_schizophrenia_dx1_all_time
## Min. :0.0000 Min. :0.00000
## 1st Qu.:0.0000 1st Qu.:0.00000
## Median :0.0000 Median :0.00000
## Mean :0.1174 Mean :0.01215
## 3rd Qu.:0.0000 3rd Qu.:0.00000
## Max. :2.0000 Max. :1.00000
##
## readmit_flag los
## Min. :0.000 Min. : 0.000
## 1st Qu.:0.000 1st Qu.: 1.000
## Median :0.000 Median : 2.000
## Mean :0.332 Mean : 3.255
## 3rd Qu.:1.000 3rd Qu.: 3.000
## Max. :1.000 Max. :98.000
## 7. Readmission rates by age quartile
# load functions:
source(here::here("code",
"readmit-rates-by-age_function.R"))
# map function across nested df:
df4.site.nested %<>%
mutate(readmit_rates = map(data,
readmit_rates_function,
exclude_los_zero = 1),
# now add the column plots:
readmit_plot = pmap(list(readmit_rates,
as.character(institution),
overall_readmit),
readmit_rates_plot))
# results:
df4.site.nested$readmit_plot[[1]]
df4.site.nested$readmit_plot## [[1]]
##
## [[2]]
##
## [[3]]
##
## [[4]]
##
## [[5]]
##
## [[6]]
##
## [[7]]
##
## [[8]]
##
## [[9]]
##
## [[10]]
Notes
- Recall the effects of Age and LOS are correlated: it looks like people in age quantile 4 have lowest readmit %, but this is not necessarily the effect of age alone: see
df4.site.nested$los_boxplot- the oldest patients also have longest LOS * So maybe it’s the long LOS that’s causing them to have low readmits, not the fact that they are older. - … Actually, it’s both, which is why both variables should be in the model.
# save results:
# pdf(here::here("results",
# "code output",
# "2019-02-13_readmit-rates-by-age-quantile.pdf"),
# width = 10)
# df4.site.nested$readmit_plot
# dev.off()df4.site.nested %>%
unnest(readmit_rates) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped")) %>%
scroll_box(width = "100%", height = "400px")| institution | overall_readmit | age_quantile | cases | readmit_cases | prop | exclude_los_zero | last_2_years |
|---|---|---|---|---|---|---|---|
| VGH | 0.1794504 | 0 | 11 | 2 | 0.1818182 | 1 | 1 |
| VGH | 0.1794504 | 1 | 964 | 162 | 0.1680498 | 1 | 1 |
| VGH | 0.1794504 | 2 | 1095 | 200 | 0.1826484 | 1 | 1 |
| VGH | 0.1794504 | 3 | 1063 | 235 | 0.2210724 | 1 | 1 |
| VGH | 0.1794504 | 4 | 1052 | 152 | 0.1444867 | 1 | 1 |
| UBCH | 0.0828804 | 0 | 6 | 0 | 0.0000000 | 1 | 1 |
| UBCH | 0.0828804 | 1 | 178 | 10 | 0.0561798 | 1 | 1 |
| UBCH | 0.0828804 | 2 | 175 | 18 | 0.1028571 | 1 | 1 |
| UBCH | 0.0828804 | 3 | 191 | 23 | 0.1204188 | 1 | 1 |
| UBCH | 0.0828804 | 4 | 186 | 10 | 0.0537634 | 1 | 1 |
| SPH | 0.1881597 | 0 | 12 | 3 | 0.2500000 | 1 | 1 |
| SPH | 0.1881597 | 1 | 1066 | 235 | 0.2204503 | 1 | 1 |
| SPH | 0.1881597 | 2 | 1079 | 239 | 0.2215014 | 1 | 1 |
| SPH | 0.1881597 | 3 | 1103 | 199 | 0.1804170 | 1 | 1 |
| SPH | 0.1881597 | 4 | 1098 | 144 | 0.1311475 | 1 | 1 |
| LGH | 0.1189904 | 0 | 8 | 0 | 0.0000000 | 1 | 1 |
| LGH | 0.1189904 | 1 | 402 | 40 | 0.0995025 | 1 | 1 |
| LGH | 0.1189904 | 2 | 425 | 54 | 0.1270588 | 1 | 1 |
| LGH | 0.1189904 | 3 | 399 | 61 | 0.1528822 | 1 | 1 |
| LGH | 0.1189904 | 4 | 430 | 43 | 0.1000000 | 1 | 1 |
| RHS | 0.1450428 | 0 | 3 | 0 | 0.0000000 | 1 | 1 |
| RHS | 0.1450428 | 1 | 373 | 44 | 0.1179625 | 1 | 1 |
| RHS | 0.1450428 | 2 | 405 | 73 | 0.1802469 | 1 | 1 |
| RHS | 0.1450428 | 3 | 422 | 53 | 0.1255924 | 1 | 1 |
| RHS | 0.1450428 | 4 | 431 | 67 | 0.1554524 | 1 | 1 |
| PRGH | 0.1423841 | 0 | 2 | 0 | 0.0000000 | 1 | 1 |
| PRGH | 0.1423841 | 1 | 71 | 18 | 0.2535211 | 1 | 1 |
| PRGH | 0.1423841 | 2 | 73 | 8 | 0.1095890 | 1 | 1 |
| PRGH | 0.1423841 | 3 | 77 | 9 | 0.1168831 | 1 | 1 |
| PRGH | 0.1423841 | 4 | 79 | 8 | 0.1012658 | 1 | 1 |
| SMH | 0.1646859 | 0 | 2 | 0 | 0.0000000 | 1 | 1 |
| SMH | 0.1646859 | 1 | 146 | 20 | 0.1369863 | 1 | 1 |
| SMH | 0.1646859 | 2 | 143 | 39 | 0.2727273 | 1 | 1 |
| SMH | 0.1646859 | 3 | 144 | 22 | 0.1527778 | 1 | 1 |
| SMH | 0.1646859 | 4 | 154 | 16 | 0.1038961 | 1 | 1 |
| SGH | 0.1038961 | 1 | 18 | 3 | 0.1666667 | 1 | 1 |
| SGH | 0.1038961 | 2 | 19 | 3 | 0.1578947 | 1 | 1 |
| SGH | 0.1038961 | 3 | 19 | 0 | 0.0000000 | 1 | 1 |
| SGH | 0.1038961 | 4 | 21 | 2 | 0.0952381 | 1 | 1 |
| RWLMH | 0.0303030 | 0 | 33 | 1 | 0.0303030 | 1 | 1 |
| BCGH | 0.3670886 | 0 | 79 | 29 | 0.3670886 | 1 | 1 |
8. Readmission rates by LOS decile
# load functions:
source(here::here("code",
"readmit-rates-by-los_function.R"))
# map function across nested df:
df4.site.nested %<>%
mutate(readmit_rates_by_los = map(data,
readmit_rates_los_function,
exclude_los_zero = 1),
readmit_by_los_plot = pmap(list(readmit_rates_by_los,
as.character(institution),
overall_readmit),
readmit_rates_by_los_plot))
# results:
# df4.site.nested$readmit_rates_by_los
# df4.site.nested$readmit_by_los_plot[[2]]
df4.site.nested$readmit_by_los_plot## [[1]]
##
## [[2]]
##
## [[3]]
##
## [[4]]
##
## [[5]]
##
## [[6]]
##
## [[7]]
##
## [[8]]
##
## [[9]]
##
## [[10]]
# write output:
# pdf(here::here("results",
# "code output",
# "2019-02-13_readmit-rates-by-LOS-decile.pdf"),
# width = 10)
# df4.site.nested$readmit_by_los_plot
# dev.off()9. Readmission rates by Gender
# load functions:
source(here::here("code",
"readmit-rates-by-gender_function.R"))
# map function across nested df:
df4.site.nested %<>%
mutate(readmit_rates_by_gender = map(data,
readmit_rates_gender_function,
exclude_los_zero = 1),
readmit_plot_by_gender = pmap(list(readmit_rates_by_gender,
as.character(institution),
overall_readmit),
readmit_rates_by_gender_plot)
)
# result:
df4.site.nested %>%
unnest(readmit_rates_by_gender) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped")) %>%
scroll_box(width = "100%", height = "400px")| institution | overall_readmit | gender | cases | readmit_cases | prop | exclude_los_zero | last_2_years |
|---|---|---|---|---|---|---|---|
| VGH | 0.1794504 | F | 1790 | 325 | 0.1815642 | 1 | 1 |
| VGH | 0.1794504 | M | 2384 | 424 | 0.1778523 | 1 | 1 |
| VGH | 0.1794504 | NULL | 11 | 2 | 0.1818182 | 1 | 1 |
| UBCH | 0.0828804 | F | 420 | 28 | 0.0666667 | 1 | 1 |
| UBCH | 0.0828804 | M | 310 | 33 | 0.1064516 | 1 | 1 |
| UBCH | 0.0828804 | NULL | 6 | 0 | 0.0000000 | 1 | 1 |
| SPH | 0.1881597 | F | 1616 | 262 | 0.1621287 | 1 | 1 |
| SPH | 0.1881597 | M | 2730 | 555 | 0.2032967 | 1 | 1 |
| SPH | 0.1881597 | NULL | 12 | 3 | 0.2500000 | 1 | 1 |
| LGH | 0.1189904 | F | 835 | 109 | 0.1305389 | 1 | 1 |
| LGH | 0.1189904 | M | 821 | 89 | 0.1084044 | 1 | 1 |
| LGH | 0.1189904 | NULL | 8 | 0 | 0.0000000 | 1 | 1 |
| RHS | 0.1450428 | F | 672 | 90 | 0.1339286 | 1 | 1 |
| RHS | 0.1450428 | M | 959 | 147 | 0.1532847 | 1 | 1 |
| RHS | 0.1450428 | NULL | 3 | 0 | 0.0000000 | 1 | 1 |
| PRGH | 0.1423841 | F | 159 | 25 | 0.1572327 | 1 | 1 |
| PRGH | 0.1423841 | M | 141 | 18 | 0.1276596 | 1 | 1 |
| PRGH | 0.1423841 | NULL | 2 | 0 | 0.0000000 | 1 | 1 |
| SMH | 0.1646859 | F | 314 | 53 | 0.1687898 | 1 | 1 |
| SMH | 0.1646859 | M | 273 | 44 | 0.1611722 | 1 | 1 |
| SMH | 0.1646859 | NULL | 2 | 0 | 0.0000000 | 1 | 1 |
| SGH | 0.1038961 | F | 43 | 5 | 0.1162791 | 1 | 1 |
| SGH | 0.1038961 | M | 34 | 3 | 0.0882353 | 1 | 1 |
| RWLMH | 0.0303030 | NULL | 33 | 1 | 0.0303030 | 1 | 1 |
| BCGH | 0.3670886 | NULL | 79 | 29 | 0.3670886 | 1 | 1 |
# graphs:
df4.site.nested$readmit_plot_by_gender## [[1]]
##
## [[2]]
##
## [[3]]
##
## [[4]]
##
## [[5]]
##
## [[6]]
##
## [[7]]
##
## [[8]]
##
## [[9]]
##
## [[10]]
# write output:
# pdf(here::here("results",
# "code output",
# "2019-02-13_readmit-rates-by-gender.pdf"),
# width = 10)
# df4.site.nested$readmit_plot_by_gender
# dev.off()10. Pull nested data together to print
df5.0.overall.readmits <-
df4.site.nested %>%
unnest(readmission_overall) %>%
select(institution,
cases,
readmit_cases,
prop)
df5.1.age.readmits <-
df4.site.nested %>%
unnest(readmit_rates) %>%
select(institution,
age_quantile,
prop) %>%
spread(key = age_quantile,
value = prop)
df5.2.los.readmits <-
df4.site.nested %>%
unnest(readmit_rates_by_los) %>%
select(institution,
los_decile,
prop) %>%
spread(key = los_decile,
value = prop) %>%
select(-`10`,
`10`)
df5.3.gender.readmits <-
df4.site.nested %>%
unnest(readmit_rates_by_gender) %>%
select(institution,
gender,
prop) %>%
spread(key = gender,
value = prop)
# join them all together:
df5.summary <-
df5.0.overall.readmits %>%
inner_join(df5.1.age.readmits,
by = c("institution" = "institution")) %>%
inner_join(df5.2.los.readmits,
by = c("institution" = "institution")) %>%
inner_join(df5.3.gender.readmits,
by = c("institution" = "institution")) %>%
purrr::set_names(c("institution",
"overall_cases",
"overall_readmits",
"overall_prop",
"age_quantile_0",
"age_quantile_1",
"age_quantile_2",
"age_quantile_3",
"age_quantile_4",
"los_decile_1",
"los_decile_2",
"los_decile_3",
"los_decile_4",
"los_decile_5",
"los_decile_6",
"los_decile_7",
"los_decile_8",
"los_decile_9",
"los_decile_10",
"gender_female",
"gender_male",
"gender_NULL"))
# result
df5.summary %>%
kable() %>%
kable_styling(bootstrap_options = c("striped")) %>%
scroll_box(width = "100%", height = "400px")| institution | overall_cases | overall_readmits | overall_prop | age_quantile_0 | age_quantile_1 | age_quantile_2 | age_quantile_3 | age_quantile_4 | los_decile_1 | los_decile_2 | los_decile_3 | los_decile_4 | los_decile_5 | los_decile_6 | los_decile_7 | los_decile_8 | los_decile_9 | los_decile_10 | gender_female | gender_male | gender_NULL |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| VGH | 4185 | 751 | 0.1794504 | 0.1818182 | 0.1680498 | 0.1826484 | 0.2210724 | 0.1444867 | NA | 0.1807229 | 0.2176871 | 0.2061657 | 0.2202073 | 0.1953782 | 0.2019002 | 0.1789976 | 0.1294964 | 0.0898345 | 0.1815642 | 0.1778523 | 0.1818182 |
| UBCH | 736 | 61 | 0.0828804 | 0.0000000 | 0.0561798 | 0.1028571 | 0.1204188 | 0.0537634 | 0.1944444 | 0.0769231 | 0.1071429 | 0.0428571 | 0.0649351 | 0.0821918 | 0.0625000 | 0.0641026 | 0.0253165 | 0.1081081 | 0.0666667 | 0.1064516 | 0.0000000 |
| SPH | 4358 | 820 | 0.1881597 | 0.2500000 | 0.2204503 | 0.2215014 | 0.1804170 | 0.1311475 | NA | 0.1869031 | 0.1917211 | 0.2033582 | 0.2556818 | 0.2490637 | 0.1818182 | 0.1698925 | 0.1341176 | 0.1185682 | 0.1621287 | 0.2032967 | 0.2500000 |
| LGH | 1664 | 198 | 0.1189904 | 0.0000000 | 0.0995025 | 0.1270588 | 0.1528822 | 0.1000000 | NA | 0.1446281 | 0.1398964 | 0.0970149 | 0.1336406 | 0.1386139 | 0.1034483 | 0.0943396 | 0.1028571 | 0.0892857 | 0.1305389 | 0.1084044 | 0.0000000 |
| RHS | 1634 | 237 | 0.1450428 | 0.0000000 | 0.1179625 | 0.1802469 | 0.1255924 | 0.1554524 | NA | NA | 0.1343284 | NA | 0.1554054 | 0.1621622 | 0.1681416 | 0.1272727 | 0.1428571 | 0.1301775 | 0.1339286 | 0.1532847 | 0.0000000 |
| PRGH | 302 | 43 | 0.1423841 | 0.0000000 | 0.2535211 | 0.1095890 | 0.1168831 | 0.1012658 | NA | 0.1754386 | 0.1200000 | 0.1666667 | 0.0952381 | 0.1200000 | 0.1250000 | 0.1333333 | 0.1142857 | 0.2187500 | 0.1572327 | 0.1276596 | 0.0000000 |
| SMH | 589 | 97 | 0.1646859 | 0.0000000 | 0.1369863 | 0.2727273 | 0.1527778 | 0.1038961 | NA | 0.2500000 | NA | 0.1904762 | 0.1333333 | 0.1346154 | 0.1578947 | 0.1269841 | 0.1355932 | 0.1093750 | 0.1687898 | 0.1611722 | 0.0000000 |
| SGH | 77 | 8 | 0.1038961 | NA | 0.1666667 | 0.1578947 | 0.0000000 | 0.0952381 | NA | 0.2000000 | NA | NA | 0.1363636 | NA | 0.0833333 | 0.0909091 | 0.0000000 | 0.0000000 | 0.1162791 | 0.0882353 | NA |
| RWLMH | 33 | 1 | 0.0303030 | 0.0303030 | NA | NA | NA | NA | NA | NA | NA | NA | 0.0000000 | NA | NA | 0.0000000 | 0.2500000 | 0.0000000 | NA | NA | 0.0303030 |
| BCGH | 79 | 29 | 0.3670886 | 0.3670886 | NA | NA | NA | NA | NA | NA | 0.3888889 | NA | NA | 0.3928571 | NA | 0.3529412 | 0.4285714 | 0.2222222 | NA | NA | 0.3670886 |
# WRITE OUTPUT:
write_csv(df5.summary,
here::here("results",
"code output",
"2019-02-13_summary-readmit-rates.csv"))