This reads in the NSQIP data for all shunt cases and all peds cases from 2019. Next, we find all the unique cases of shunt placement and merges with record data. Then, we create our final working dataframe that only contains the factors of interest.
everything above this line is executed locally/on desktop. The below code uses uploaded CSV created on desktop.
Pull out the columns of interest.
NSQIP Cohort A: 1) patient received first permanent shunt 2) patient has a recorded hydrocephalus etiology 3) patient has a recorded entry for premature birth (excl UNKNOWN) 4) patient did not die in first 30 days
1676 meet the criteria
cohortA <- df19[which(df19$PED_SHNT_1STPERM == "Yes"),]
cohortA <- cohortA[which(!cohortA$PREM_BIRTH == "Unknown"),]
cohortA <- cohortA[which(!cohortA$DEATH30YN == "Yes"),]
df <- cohortAHere, we replace “-99” gestational age with “39” if the baby was not born premature. The reason being that when we initially did this analysis, ages of patients with a reported GESTATIONALAGE_BIRTH were strongly skewed premature. This was determined to be because this category was generally only reported for patients who were born premature. To remediate this, ages were interpolated by finding the known median age of full term birth patients and assigning that age to all full term births known by the category PREM_BIRTH. The median age was found to be 39 weeks.
#first find median age of existing full term births to interpolate missing values
median(df$GESTATIONALAGE_BIRTH[df$GESTATIONALAGE_BIRTH>36])## [1] 39#because the median age of a full term birth from existing data is 39, we interpolate 39 for all cases of a full term before with no listed GA at birth
df$GESTATIONALAGE_BIRTH[df$PREM_BIRTH=="No" & df$GESTATIONALAGE_BIRTH==-99] = 39
df$GESTATIONALAGE_BIRTH[df$GESTATIONALAGE_BIRTH==-99] <- NAAdd category for chronological age in weeks
(CHRONAGE_SURGERY) by dividing age at surgery by 7 and
rounding to the nearest whole number.
Calculate gestational age at surgery by adding gestational age at birth plus chronological age at surgery.
df$CHRONAGE_SURGERY <- round(df$AGE_DAYS/7,0)
df$GESTATIONALAGE_SURGERY2 <- df$GESTATIONALAGE_BIRTH+df$CHRONAGE_SURGERYHispanic/ethnicity
All Races All etiologies (everything that is not UNC is other)
-UNC categories are Congenital hydrocephalus Intraventricular hemorrhage
(IVH) or prematurity or without cause, Neoplastic Post infectious , Post
Traumatic , Spin bifida with Chiari malformation/myelomenigocele
Syndromic, not otherwise specified
-ONLY include prior procedure in Premie IVH table -premature birth is YES/NO (less than 37 weeks)
descriptive stats
df[,c(3:5,9,13,14,8)] %>%
mutate(PREM_BIRTH = ifelse(df$PREM_BIRTH=="No", "No", "Yes")) %>%
mutate(PED_SHNT_HC_ETIOL = ifelse(df$PED_SHNT_HC_ETIOL=="Vascular anomaly" | df$PED_SHNT_HC_ETIOL=="Syndromic, not otherwise specified" | df$PED_SHNT_HC_ETIOL=="Intracranial cyst" | df$PED_SHNT_HC_ETIOL=="Post hemorrhagic from other vascular lesion" | df$PED_SHNT_HC_ETIOL == "Other - insufficient information to classify type", "Syndromic, not otherwise specified ", PED_SHNT_HC_ETIOL)) %>%
tbl_summary() | Characteristic | N = 1,6761 |
|---|---|
| SEX | |
| Female | 748 (45%) |
| Male | 928 (55%) |
| RACE | |
| American Indian or Alaska Native | 8 (0.5%) |
| Asian | 43 (2.6%) |
| Black or African American | 344 (21%) |
| Native Hawaiian or Other Pacific Islander | 4 (0.2%) |
| Unknown/Not Reported | 299 (18%) |
| White | 978 (58%) |
| ETHNICITY_HISPANIC | |
| No | 1,222 (73%) |
| NULL | 166 (9.9%) |
| Yes | 288 (17%) |
| GESTATIONALAGE_BIRTH | 38.0 (32.0, 39.0) |
| Unknown | 80 |
| PED_SHNT_HC_ETIOL | |
| Congenital hydrocephalus | 520 (31%) |
| Intraventricular hemorrhage (IVH) of prematurity without other cause | 372 (22%) |
| Neoplastic | 183 (11%) |
| Post infectious | 73 (4.4%) |
| Post traumatic | 39 (2.3%) |
| Spina bifida with Chiari malformation / myelomeningocele | 117 (7.0%) |
| Syndromic, not otherwise specified | 372 (22%) |
| PED_SHNT_PRIOR_HCPROC | 591 (35%) |
| PREM_BIRTH | 730 (44%) |
| 1 n (%); Median (IQR) | |
df %>%
filter(!GESTATIONALAGE_BIRTH == -99) %>%
summarise(GA_birth = median(GESTATIONALAGE_BIRTH), CA_surgery = median(CHRONAGE_SURGERY), GA_surgery = median(GESTATIONALAGE_SURGERY2))## GA_birth CA_surgery GA_surgery
## 1 38 16 49#most frequent etiologies
df %>%
group_by(PED_SHNT_HC_ETIOL) %>%
dplyr::summarise(n = n()) %>%
arrange(desc(n))## # A tibble: 11 × 2
## PED_SHNT_HC_ETIOL n
## <chr> <int>
## 1 Congenital hydrocephalus 520
## 2 Intraventricular hemorrhage (IVH) of prematurity without other cause 372
## 3 Neoplastic 183
## 4 Other - insufficient information to classify type 149
## 5 Spina bifida with Chiari malformation / myelomeningocele 117
## 6 Intracranial cyst 109
## 7 Post infectious 73
## 8 Post hemorrhagic from other vascular lesion 58
## 9 Post traumatic 39
## 10 Syndromic, not otherwise specified 37
## 11 Vascular anomaly 19520/nrow(df)*100## [1] 31.02625372/nrow(df)*100## [1] 22.1957183/nrow(df)*100## [1] 10.91885117/nrow(df)*100 #spina bifida## [1] 6.980907race breakdown for cohort A
df %>%
group_by(RACE) %>%
dplyr::summarise(n = n()) %>%
arrange(desc(n))%>%
mutate(percentage = n / sum(n)*100)## # A tibble: 6 × 3
## RACE n percentage
## <chr> <int> <dbl>
## 1 White 978 58.4
## 2 Black or African American 344 20.5
## 3 Unknown/Not Reported 299 17.8
## 4 Asian 43 2.57
## 5 American Indian or Alaska Native 8 0.477
## 6 Native Hawaiian or Other Pacific Islander 4 0.239#premature births
df %>%
group_by(PREM_BIRTH) %>%
dplyr::summarise(n = n()) %>%
arrange(desc(n))%>%
mutate(percentage = n / sum(n)*100)## # A tibble: 9 × 3
## PREM_BIRTH n percentage
## <chr> <int> <dbl>
## 1 No 946 56.4
## 2 35-36 completed weeks gestation 156 9.31
## 3 25-26 completed weeks gestation 125 7.46
## 4 33-34 completed weeks gestation 103 6.15
## 5 27-28 completed weeks gestation 99 5.91
## 6 31-32 completed weeks gestation 65 3.88
## 7 29-30 completed weeks gestation 64 3.82
## 8 24 completed weeks gestation 62 3.70
## 9 Less than 24 completed weeks gestation 56 3.34print("Percentage of patients born premature:")## [1] "Percentage of patients born premature:"nrow(cohortA[which(!cohortA$PREM_BIRTH == "No"),])/nrow(cohortA)*100## [1] 43.55609#shunt failure within 30 days
df %>%
filter(!PED_SHNT_INTRE1_DAYS == -99) %>%
summarise(med_days_to_fail = median(PED_SHNT_INTRE1_DAYS))## med_days_to_fail
## 1 17#median time to failure 17 days for the 152 of cases (9.069212%) that failed within 30 days
nrow(cohortA[which(!cohortA$PED_SHNT_INTRE1_DAYS == -99),])## [1] 152nrow(cohortA[which(!cohortA$PED_SHNT_INTRE1_DAYS == -99),])/nrow(cohortA)*100## [1] 9.069212#Causes of shunt failure
df %>%
filter(!PED_SHNT_INTRE1_RSN == "NULL") %>%
group_by(PED_SHNT_INTRE1_RSN) %>%
dplyr::summarise(n = n()) %>%
arrange(desc(n))## # A tibble: 7 × 2
## PED_SHNT_INTRE1_RSN n
## <chr> <int>
## 1 Shunt infection 48
## 2 Wound disruption or CSF leak 36
## 3 Mechanical issues 21
## 4 Other 17
## 5 Proximal catheter obstruction or malposition 15
## 6 Distal catheter obstruction or malposition 11
## 7 Loculated ventricle 6#Most common reported causes: Shunt infection (n=48), Wound Disruption or CSF Leak (n=36), and Mechanical issues (n=21).
48/152*100## [1] 31.5789536/152*100## [1] 23.6842121/152*100## [1] 13.81579#Cases for which we have gestational age at surgery (1596)
df %>%
filter(!GESTATIONALAGE_BIRTH == -99) %>%
dplyr::summarise(n = n()) ## n
## 1 1596#cases that had their surgery BEFORE 40 weeks of gestational age (407)
df_before40 <- df %>%
filter(!GESTATIONALAGE_BIRTH == -99) %>%
filter(GESTATIONALAGE_SURGERY2 < 40)
#cases that had their surgery AT OR AFTER 40 weeks of gestational age (1189)
df_after40 <- df %>%
filter(!GESTATIONALAGE_BIRTH == -99) %>%
filter(GESTATIONALAGE_SURGERY2 > 39) NSQIP Cohort A: Number of patients who had a shunt placed after 40 GA: 1189 # born prematurely (<37 weeks aka 259 days): 346 Failure rate in this population: 10.7% (n=37) # born full term (>37 weeks aka 260+ days): 843 Failure rate in this population: 7.6% (n=64)
#surgery was after 40 weeks GA AND born premature
df_after40 %>%
filter(GESTATIONALAGE_BIRTH <37)%>%
dplyr::summarise(n = n()) ## n
## 1 346after40_prem_fail <- df_after40 %>%
filter(GESTATIONALAGE_BIRTH <37)%>%
filter(!PED_SHNT_INTRE1 == "None")%>%
dplyr::summarise(n = n())
after40_prem_fail/346*100## n
## 1 10.69364#surgery was after 40 weeks GA AND born full term
df_after40 %>%
filter(GESTATIONALAGE_BIRTH >36)%>%
dplyr::summarise(n = n()) ## n
## 1 843after40_full_fail <- df_after40 %>%
filter(GESTATIONALAGE_BIRTH >36)%>%
filter(!PED_SHNT_INTRE1 == "None")%>%
dplyr::summarise(n = n())
after40_full_fail/843*100## n
## 1 7.591934Failures for under/over 40 wks groups
#47 failures in the before40 weeks op group
df_before40 %>%
filter(!PED_SHNT_INTRE1_DAYS == -99) %>%
dplyr::summarise(n = n()) ## n
## 1 47#100 failures in the after40 weeks op group
df_after40 %>%
filter(!PED_SHNT_INTRE1_DAYS == -99) %>%
dplyr::summarise(n = n()) ## n
## 1 100#Before 40 has a 12% failure; after 40 has an 8% failure
47/407*100;100/1189*100## [1] 11.54791## [1] 8.410429#Median of 17 days to failure in before40 group
df_before40 %>%
filter(!PED_SHNT_INTRE1_DAYS == -99) %>%
summarise(med_days_to_fail = median(PED_SHNT_INTRE1_DAYS))## med_days_to_fail
## 1 17#Median of 17 days to failure in before40 group
df_after40 %>%
filter(!PED_SHNT_INTRE1_DAYS == -99) %>%
summarise(med_days_to_fail = median(PED_SHNT_INTRE1_DAYS))## med_days_to_fail
## 1 17make new df of failed and non failed cases
#cases that had their shunt FAIL in the first 30 days (n=147)
df_fail<- df %>%
filter(!GESTATIONALAGE_BIRTH == -99) %>%
filter(!PED_SHNT_INTRE1_DAYS == -99)
#cases that did not have a shunt failure in the first 30 days (n=1449)
df_ok <- df %>%
filter(!GESTATIONALAGE_BIRTH == -99) %>%
filter(PED_SHNT_INTRE1_DAYS == -99) fishers exact predicting if pre-maturity predicts shunt failure
table(df_ok$prem_cat)## < table of extent 0 >table(df_fail$prem_cat)## < table of extent 0 >577/(577+872)## [1] 0.398205774/(74+73)## [1] 0.5034014df_ok$prem_cat <- ifelse(df_ok$PREM_BIRTH=="No","Not Prem", "Premature")
df_fail$prem_cat <- ifelse(df_fail$PREM_BIRTH=="No","Not Prem", "Premature")#<40 week surgery: y = 47, n = 360
#>40 week surgery: y = 100, n = 1089wilcox test of age between failure and non failure groups.
wilcox.test(df_fail$GESTATIONALAGE_BIRTH,df_ok$GESTATIONALAGE_BIRTH)##
## Wilcoxon rank sum test with continuity correction
##
## data: df_fail$GESTATIONALAGE_BIRTH and df_ok$GESTATIONALAGE_BIRTH
## W = 98240, p-value = 0.1136
## alternative hypothesis: true location shift is not equal to 0wilcox.test(df_fail$GESTATIONALAGE_SURGERY2,df_ok$GESTATIONALAGE_SURGERY2)##
## Wilcoxon rank sum test with continuity correction
##
## data: df_fail$GESTATIONALAGE_SURGERY2 and df_ok$GESTATIONALAGE_SURGERY2
## W = 89724, p-value = 0.00162
## alternative hypothesis: true location shift is not equal to 0wilcox.test(df_fail$CHRONAGE_SURGERY,df_ok$CHRONAGE_SURGERY)##
## Wilcoxon rank sum test with continuity correction
##
## data: df_fail$CHRONAGE_SURGERY and df_ok$CHRONAGE_SURGERY
## W = 89136, p-value = 0.001104
## alternative hypothesis: true location shift is not equal to 0For this analysis we need the patients with hydrocephalus etiology of only Intraventricular Hemorrhage AND Born Prematurely (which I am calling NSQIP Cohort B). To identify this group in the NSQIP dataset - they have “Intraventricular hemorrhage (IVH) or prematurity or without cause” (PED_SHNT_HC_ETIOL) AND were born <37 weeks gestational age aka yes born pre-maturely in the (PREM_BIRTH) variable))
#unique(df19$PED_SHNT_HC_ETIOL)
IVH <- df[which(df$PED_SHNT_HC_ETIOL == "Intraventricular hemorrhage (IVH) of prematurity without other cause"),]
cat("Number of patients with Intraventricular hemorrhage (IVH) of prematurity without other cause:",nrow(IVH)) #372## Number of patients with Intraventricular hemorrhage (IVH) of prematurity without other cause: 372Within IVH subset, how many are actually prematurely born: Within IVH category, # of patients with (PREM_BIRTH) variable any input other than “no” or “unknown” = ________
cohortB <- IVH[which(!IVH$PREM_BIRTH=="No"),]
#cohortB <- IVH[which(IVH$GESTATIONALAGE_BIRTH <37),]
cohortB <- cohortB[which(!cohortB$GESTATIONALAGE_BIRTH == -99),]
cat("\nnumber of patients prematurely born within IVH subset:",nrow(cohortB)) #321##
## number of patients prematurely born within IVH subset: 321df <- cohortB
#list of the different prem_birth categories and frequencies
df %>%
group_by(PREM_BIRTH) %>%
dplyr::summarise(n = n()) %>%
arrange(desc(n))## # A tibble: 8 × 2
## PREM_BIRTH n
## <chr> <int>
## 1 25-26 completed weeks gestation 83
## 2 27-28 completed weeks gestation 57
## 3 24 completed weeks gestation 51
## 4 Less than 24 completed weeks gestation 42
## 5 35-36 completed weeks gestation 27
## 6 29-30 completed weeks gestation 26
## 7 31-32 completed weeks gestation 19
## 8 33-34 completed weeks gestation 16Now we have NSQIP Cohort B (patients with both IVH and were born prematurely)
Within NSQIP Cohort B: how many had shunt implant surgery before 40 weeks gestational age (this is that calculated variable): ________ how many had shunt implant surgery after 40 weeks gestational age (this is that calculated variable): ________
#IVH cases that had their surgery BEFORE 40 weeks of gestational age
ivh_before40 <- df %>%
filter(!GESTATIONALAGE_SURGERY2 < 40)
cat("number of IVH patients with shunt implant surgery before 40 weeks gestational age:",nrow(ivh_before40)) #163## number of IVH patients with shunt implant surgery before 40 weeks gestational age: 157#IVH cases that had their surgery AT OR AFTER 40 weeks of gestational age
ivh_after40 <- df %>%
filter(!GESTATIONALAGE_SURGERY2 > 39)
cat("\nnumber of IVH patients with shunt implant surgery at or after 40 weeks gestational age:",nrow(ivh_after40)) #179##
## number of IVH patients with shunt implant surgery at or after 40 weeks gestational age: 164Split NSQIP Cohort B into Before40Weeks and After40Weeks
In Before40Weeks gestational age surgery group: How many had a prior hydrocephalus procedure: ______ Variable “PED_SHNT_PRIOR_HCPROC” (yes, no), we want “yes”
ivh_hydro <- ivh_before40[which(ivh_before40$PED_SHNT_PRIOR_HCPROC == "Yes"),]
cat("\nnumber of IVH patients w surgery before 40 wks with prior hydro procedure:",nrow(ivh_hydro)) #103##
## number of IVH patients w surgery before 40 wks with prior hydro procedure: 102How many had a shunt failure: ______ Variable “PED_SHNT_INTRE1”, any entry other than “none”
ivh_fail<- ivh_before40[which(!ivh_before40$PED_SHNT_INTRE1_RSN == "NULL"),]
cat("\nnumber of IVH patients w surgery before 40 wks with shunt failure:",nrow(ivh_fail)) #18##
## number of IVH patients w surgery before 40 wks with shunt failure: 18#calculated differently
ivh_fail2 <- ivh_before40[which(!ivh_before40$PED_SHNT_INTRE1 == "None"),]
cat("/nnumber of IVH patients w surgery before 40 weeks with shunt failure:",nrow(ivh_fail2))## /nnumber of IVH patients w surgery before 40 weeks with shunt failure: 18In after 40 weeks gestational age surgery group: How many had a prior hydrocephalus procedure: ______
ivh_hydro <- ivh_after40[which(ivh_after40$PED_SHNT_PRIOR_HCPROC == "Yes"),]
cat("\nnumber of IVH patients w surgery after 40 wks with prev hydro procedure:",nrow(ivh_hydro)) #113##
## number of IVH patients w surgery after 40 wks with prev hydro procedure: 106How many had a shunt failure: ______ Variable “PED_SHNT_INTRE1”, any entry other than “none”
ivh_fail<- ivh_after40[which(!ivh_after40$PED_SHNT_INTRE1_RSN == "NULL"),]
cat("\nnumber of IVH patients w surgery after 40 wks with shunt failure:",nrow(ivh_fail)) #19##
## number of IVH patients w surgery after 40 wks with shunt failure: 17number of IVH patients with shunt implant surgery before 40 weeks gestational age: number of IVH patients w surgery before 40 wks with shunt failure:
cat("Percentage of shunt failures in IVH patients w surgery BEFORE 40 weeks:")## Percentage of shunt failures in IVH patients w surgery BEFORE 40 weeks:number of IVH patients with shunt implant surgery at or after 40 weeks gestational age: number of IVH patients w surgery after 40 wks with shunt failure:
cat("\nPercentage of shunt failures in IVH patients w surgery AT/AFTER 40 weeks:")##
## Percentage of shunt failures in IVH patients w surgery AT/AFTER 40 weeks: