vitals_feb24

Methods

KHCC cancer registry was quieried for all patients
Vitals for these patients were downloaded from our Vista software using fileman
The 2 datasets were merged and used for downstream analysis
R version 4.3.2 was used
Childsds package was used – link to this package: https://cran.r-project.org/web/packages/childsds/index.html – Reference: use website link as a referene and list the 3 references below to refer to Turkish growth charts
We calculated the time in months since diagnosis for the measurement and cut this into cateogries. Weight and Height per year after diagnosis were averaged for all reading during the year
The first reading of height and weight was used as “Diagnosis values”. First readings after 1 and 5 years following diagnosis were used.
logistic regression was used to build a prediction model for the chances of having obesity and short stature at 5 years.
To find the common values of our patients we used an arbitrary value of 10% and 90%. WE did not perform parametric analysis with standard deviation and z scores as data was not normally distributed.
BMI definitions: <5% underweight, 5-85% health, 85-95% overweight >95% obeses
Hatipoglu, N. et al. Waist circumference percentiles for 7- to 17-year-old Turkish children and adolescents. Eur J Pediatr 167, 383–389 (2008);Bundak, R. et al. Body mass index references for Turkish children. Acta Paediatrica 95, 194–198 (2006).
Neyzi, O., Furman, A., Bundak, R., Gunoz, H., Darendeliler, F., Bas, F., 2006. Growth references for Turkish children aged 6 to 18 years. Acta Paediatrica 95, 1635-1641. doi:10.1080/08035250600652013
Bundak, R. et al. Body mass index references for Turkish children. Acta Paediatrica 95, 194-198 (2006).

lookupCategory<-read.csv("lookupCAtegory.csv")


ds<- read.csv("list19Apr23.csv") %>% 
  clean_names() %>% 
  mutate(across(where(is.character), stringr::str_to_upper)) %>%    
  mutate(across(where(is.character), stringr::str_squish))  %>%     
  mutate(date_of_death=if_else(date_of_death=="00/00/0000", "4/19/2023", date_of_death)) %>% 
  mutate(across(contains(c("date", "dte")), function(x) parse_date_time2(x, order=c("dBY" , "mdY", "dbY")))) %>% 
  mutate(dead=if_else(vital_status== "ALIVE(1)", 0, 1)) %>% 
  mutate(OS= as.numeric(difftime(date_of_death , date_of_diagnosis, units="days"))/30) %>% 
  mutate(date_of_birth=date_of_diagnosis-years(age_at_diagnosis)) %>% 
  mutate(nationality = ifelse(nationality=="JORDAN(01)", "Jordanian", "Nonjordanian")) %>% 
  mutate(site=str_extract(primary_site, "(?<=\\()([^)]+)(?=\\))")) %>% 
  mutate(histology = str_extract(histology_behavior_icd_o_3, "\\(\\d{4}"),
         histology = str_remove(histology, "\\(")) %>% 
  mutate(record_number=  str_extract(accession_seq, "(?<=/)\\d+")) %>% 
  mutate(year_of_diagnosis=year(date_of_diagnosis)) %>% 
  filter(year_of_diagnosis>=2010) 
  

library(stringr)

# https://seer.cancer.gov/ayarecode/aya-who2008.html       4/4/2021
aya <- read.csv("ayarecodewho2008.txt", sep=";")

aya<- aya[-98,]
for (l in 1:nrow(aya)) if (aya[l,1]=="") aya[l,1]<-aya[l-1,1]

split<-function(x){
  result<-c()
  t<-unlist(strsplit(x, split=","))
  for (b in t){
    sapply(str_extract_all(b, '[0-9.]+'), function(x) as.numeric(x)) ->a
    if (length(a)>1) a<-c(a[1,1]:a[2,1])
    result<-c(result, a)}
  return (result)
}

aya$site<-list(c())
for (l in 1:nrow(aya)) if (aya[l,3]!=" ") aya$site[[l]]<-split(aya[l,3])

aya$hist<-list(c())
for (l in 1:nrow(aya)) if (aya[l,4]!=" ") aya$hist[[l]]<-split(aya[l,4])


aya$category<-c("")
for (l in 1:nrow(aya)) if (!grepl("\\.", aya[l,1])) aya$category[l]=aya[l,1]
for (l in 1:nrow(aya)) if (aya[l,8]=="") aya[l,8]<-aya[l-1,8]
excludeEmpty<-which(aya$Primary.Site==" ")
aya %>% slice(-excludeEmpty)->aya

########################################
#  patients
########################################

removeParenthesis<-function(x)gsub("\\s*\\([^\\)]+\\)","",as.character(x)) 

ds$site<- removeParenthesis(ds$site) %>%  substr(start = 2, stop = 4) %>% as.numeric
ds$histology<- ds$histology %>% as.numeric


ds$aya<-c("")
ds$ayaCategory<-c("")

for (m in 1:nrow(ds)){
  site<- ds$site[m]
  histology<- ds$histology[m]
  for (n in 1:nrow(aya)){
    if (is.element(site, aya$site[[n]]) & is.element(histology, aya$hist[[n]])){
      ds$aya[m] <-aya$SEER.AYA.Site.Recode.WHO.2008.Definition.Site.Group[n]
      ds$ayaCategory[m]<-aya$category[n]
    }}}

ds$aya<-stringr::str_squish(ds$aya)

ds<- ds%>%  left_join(lookupCategory)

ds$category<- factor(ds$category, levels=c("Leukemia", "Lymphoma", "CNS", "Solid", "Bone", "STS"))

# ds<- ds %>% filter(aya!="", OS>=0) %>% mutate(Histology=as.character(Histology))
# 
# ds$OS <- if_else(ds$Age<18 & ds$Nationality=="Jordanian" & ds$year>2015 & ds$aya=="1.1 Acute lymphoid leukemia", ds$OS*2, ds$OS)


gender<- read_csv("cancer_registry_1Feb2024.csv") %>% janitor::clean_names() %>% 
  dplyr::select(gender=sex, medical_record_number) %>% 
  distinct(medical_record_number, .keep_all = T)


ds<- left_join(ds, gender) %>% filter(!is.na(gender)) %>%  filter(!is.na(category)) %>% 
  mutate(gender=recode(gender,
                       "1 (MALE)"="MALE",
                       "2 (FEMALE)"="FEMALE"))
vitals<- read.csv("Vitals.csv") %>% 
  clean_names()%>% 
  inner_join(ds[, c("mrn", "histology_behavior_icd_o_3", "date_of_birth", "date_of_diagnosis", "category", "ayaCategory", "aya","gender", "age_at_diagnosis", "OS", "dead")]) %>% 
  mutate(date=as.Date(mdy_hm(date_time_vitals_taken))) %>% 
  mutate(age=as.numeric(difftime(date, date_of_birth,  units="days"))/360) %>% 
  mutate(months=as.numeric(difftime(date,date_of_diagnosis,  units="days"))/30)


save(ds, file="ds.RData") 
save(vitals, file="vitals.RData")

Loading data

You will get these files ds.RData and vitals.RData
Just remove the hashtags# below and load the data to have 2 dataframes ds and vitals

# load("ds.RData")
# load("vitals.RData")

HeightWeight<- vitals %>% filter(vital_type %in% c("HEIGHT", "WEIGHT"))


HeightWeight %>% 
  filter(vital_type=="WEIGHT") %>% 
  mutate( rate=as.numeric(rate), rate=0.45359237*rate) %>% 
  mutate(wfs=sds(rate, age = age, sex = gender, male = "MALE", female = "FEMALE",
                 ref = turkish.ref  , item = "weight", type = "perc")) %>% 
  dplyr::select(weight=rate, date, age, months, wfs, mrn, category, gender, age_at_diagnosis, OS, dead) -> weight



HeightWeight %>% 
  filter(vital_type=="HEIGHT") %>% 
  mutate(rate=as.numeric(rate), rate=2.54*rate) %>% 
  mutate(hfs=sds(rate, age = age, sex = gender, male = "MALE", female = "FEMALE",
                 ref = turkish.ref  , item = "height", type = "perc")) %>% 
  dplyr::select(height=rate, date, age, months, hfs, mrn, category, gender, age_at_diagnosis, OS, dead) ->height


full_join(height, weight) %>% 
  mutate(bmi=weight/((0.01*height)^2)) ->BMI


BMI<-BMI %>% mutate(bfs=sds(bmi, age = age, sex = gender, male = "MALE", female = "FEMALE",
                 ref = turkish.ref  , item = "bmi_bundak", type = "perc")) %>% 
  dplyr::select( age, months,height, hfs, weight, wfs, bmi ,  bfs,  mrn, category, gender, age_at_diagnosis, OS, dead) %>% 
  filter(age>=0, months>=0)

BMI<- BMI %>% 
  mutate(BMI_Categories = case_when(
      bfs < 0.05 ~ "Underweight",
      bfs >= 0.05 & bfs < 0.85 ~ "Healthy weight",
      bfs >= 0.85 & bfs < 0.95 ~ "Overweight",
      bfs >= 0.95 ~ "Obese",
      TRUE ~ NA_character_)) %>%  
  mutate(Height_Categories = case_when(
      hfs < 0.05 ~ "Short",
      hfs >= 0.95 ~ "Tall",
      TRUE ~ "Normal")) %>% 
mutate(Weight_Categories = case_when(
      wfs < 0.05 ~ "Low weight",
      wfs >= 0.95 ~ "Increased Weight",
      TRUE ~ "Normal"))

Donut chart showing disribution of patients

ds %>% 
  filter(mrn %in% unique(vitals$mrn)) %>% 
  mutate(var=category) %>% count(var) %>% 
  donut_chart()

Table showing charachterstics of patients

ds %>% 
  filter(mrn %in% unique(vitals$mrn)) %>% 
  mutate(year=year(date_of_diagnosis)) %>% 
  dplyr::select(c("age_at_diagnosis","gender", "year", "category", "ayaCategory")) %>% 
  tbl_summary()

Characteristic	N = 3,749¹
age_at_diagnosis	6 (2, 12)
gender
FEMALE	1,619 (43%)
MALE	2,130 (57%)
year	2,016.0 (2,014.0, 2,019.0)
category
Leukemia	1,042 (28%)
Lymphoma	619 (17%)
CNS	560 (15%)
Solid	1,065 (28%)
Bone	250 (6.7%)
STS	213 (5.7%)
ayaCategory
1 Leukemias	1,042 (28%)
10 Unspecified Malignant Neoplasms	10 (0.3%)
2 Lymphomas	619 (17%)
3 CNS and Other Intracranial and Intraspinal Neoplasms (all behaviors)	560 (15%)
4 Osseous & Chondromatous Neoplasms	250 (6.7%)
5 Soft Tissue Sarcomas	213 (5.7%)
6 Germ Cell and Trophoblastic Neoplasms	100 (2.7%)
7 Melanoma and Skin Carcinomas	15 (0.4%)
8 Carcinomas	111 (3.0%)
9 Miscellaneous specified neoplasms, NOS	829 (22%)
¹ Median (IQR); n (%)

count of patients per year of diagnosis

ds %>% 
  mutate(year=as.character(year(date_of_diagnosis))) %>% count(year) %>% gt()

year	n
2010	342
2011	384
2012	362
2013	373
2014	364
2015	384
2016	397
2017	383
2018	293
2019	336
2020	336
2021	362

median number of readings (height or weight) recorded per patient with 25 and 75% quantiles

HeightWeight %>% 
  group_by(mrn) %>% 
  summarise(n=n()) %>% 
  summarise(Total=sum(n),
            median=median(n),
    q25 = quantile(n, probs = 0.25, na.rm = TRUE),  # 25th percentile
    q75 = quantile(n, probs = 0.75, na.rm = TRUE)   # 75th percentile
  ) %>% gt()

Total	median	q25	q75
337589	66	26	130

Growth paramaters at diagnosis

Distribution lines of BMI, height and weight of children with cancer at diagnosis showing bimodal distribution of weight/BMI and mainly right-skewed curve for height

weight %>%  filter(!is.na(weight))%>%   arrange(months) %>%   group_by(mrn) %>% slice(1) -> weight_at_diagnosis
height %>%  filter(!is.na(height))%>%   arrange(months) %>%   group_by(mrn) %>% slice(1) -> height_at_diagnosis
BMI %>%     filter(!is.na(bmi))%>%      arrange(months) %>%   group_by(mrn) %>% slice(1) -> BMI_at_diagnosis



# Assuming BMI_at_diagnosis, weight_at_diagnosis, and height_at_diagnosis are your data frames
# Combine them while dropping duplicated columns

combined_df <- BMI_at_diagnosis %>%
  full_join(weight_at_diagnosis, by = intersect(names(BMI_at_diagnosis), names(weight_at_diagnosis))) %>%
  full_join(height_at_diagnosis, by = intersect(names(BMI_at_diagnosis), names(height_at_diagnosis)))

# Note: This approach joins the data frames on columns they have in common and avoids duplicating columns in the final data frame.

combined_df[, Cs(category, wfs, hfs, bfs)] %>% 
  tidyr::gather(metric, val, -category) %>% 
  ggplot(aes(x=val, fill=metric))+
  geom_density(alpha=0.3)+
  theme_classic()+
  scale_x_continuous(labels = scales::percent_format(scale = 100))+
  labs(x="Percentile for Age at Diagnosis", y="Distribution Density", fill="Metric", title="All patients combined")+
  scale_fill_discrete(labels=c("BMI", "Height", "Weight"))+
  theme(strip.background = element_blank(), axis.text.x = element_text(size=8))

combined_df[, Cs(category, wfs, hfs, bfs)] %>% 
  tidyr::gather(metric, val, -category) %>% 
  ggplot(aes(x=val, fill=metric))+
  geom_density(alpha=0.3)+
  theme_classic()+
  scale_x_continuous(labels = scales::percent_format(scale = 100))+
  labs(x="Percentile for Age at Diagnosis", y="Distribution Density", fill="Metric")+
  scale_fill_discrete(labels=c("BMI", "Height", "Weight"))+
  theme(strip.background = element_blank(), axis.text.x = element_text(size=8))+
  facet_wrap(.~category, scales="free")

growth parameters after 1 year

weight %>%  filter(!is.na(weight))%>% filter(months>=12) %>%   arrange(months) %>%   group_by(mrn) %>% slice(1) -> weight_after_1
height %>%  filter(!is.na(height))%>% filter(months>=12) %>% arrange(months) %>%   group_by(mrn) %>% slice(1) -> height_after_1
BMI %>%     filter(!is.na(bmi))%>%    filter(months>=12) %>% arrange(months) %>%   group_by(mrn) %>% slice(1) -> BMI_after_1



# Assuming BMI_after_1, weight_after_1, and height_after_1 are your data frames
# Combine them while dropping duplicated columns

combined_df <- BMI_after_1 %>%
  full_join(weight_after_1, by = intersect(names(BMI_after_1), names(weight_after_1))) %>%
  full_join(height_after_1, by = intersect(names(BMI_after_1), names(height_after_1)))

# Note: This approach j



combined_df[, Cs(category, wfs, hfs, bfs)] %>% 
  tidyr::gather(metric, val, -category) %>% 
  ggplot(aes(x=val, fill=metric))+
  geom_density(alpha=0.3)+
  theme_classic()+
  scale_x_continuous(labels = scales::percent_format(scale = 100))+
  labs(x="Percentile for Age at Diagnosis", y="Distribution Density", fill="Metric")+
  scale_fill_discrete(labels=c("BMI", "Height", "Weight"))+
  theme(strip.background = element_blank(), axis.text.x = element_text(size=8))+
  facet_wrap(.~category, scales="free")

First reading after 5 years of diagnosis

weight %>%  filter(!is.na(weight))%>% filter(months>=60) %>%   arrange(months) %>%   group_by(mrn) %>% slice(1) -> weight_after_5
height %>%  filter(!is.na(height))%>% filter(months>=60) %>% arrange(months) %>%   group_by(mrn) %>% slice(1) -> height_after_5
BMI %>%     filter(!is.na(bmi))%>%    filter(months>=60) %>% arrange(months) %>%   group_by(mrn) %>% slice(1) -> BMI_after_5



# Assuming BMI_after_5, weight_after_5, and height_after_5 are your data frames
# Combine them while dropping duplicated columns

combined_df <- BMI_after_5 %>%
  full_join(weight_after_5, by = intersect(names(BMI_after_5), names(weight_after_5))) %>%
  full_join(height_after_5, by = intersect(names(BMI_after_5), names(height_after_5)))




# Note: This approach j

combined_df[, Cs(category, wfs, hfs, bfs)] %>% 
  tidyr::gather(metric, val, -category) %>% 
  ggplot(aes(x=val, fill=metric))+
  geom_density(alpha=0.3)+
  theme_classic()+
  scale_x_continuous(labels = scales::percent_format(scale = 100))+
  labs(x="Percentile for Age at Diagnosis", y="Distribution Density", fill="Metric")+
  scale_fill_discrete(labels=c("BMI", "Height", "Weight"))+
  theme(strip.background = element_blank(), axis.text.x = element_text(size=8))+
  facet_wrap(.~category, scales="free")

table showing number of patients with abnormal values at diagnosis

BMI %>% 
  filter(!is.na(bmi)) %>% 
  arrange(months) %>% 
  group_by(mrn) %>% 
  slice(1) %>% 
  dplyr::select(category, BMI_Categories, Height_Categories, Weight_Categories ) %>% 
  tbl_summary(by=category) %>% add_overall()

Characteristic	Overall, N = 3,676¹	Leukemia, N = 1,035¹	Lymphoma, N = 611¹	CNS, N = 540¹	Solid, N = 1,042¹	Bone, N = 242¹	STS, N = 206¹
mrn	128,195 (89,131, 173,040)	131,089 (85,257, 176,207)	128,399 (91,172, 173,283)	129,276 (92,634, 173,207)	125,482 (88,878, 171,117)	128,709 (88,431, 169,525)	131,222 (94,242, 166,695)
BMI_Categories
Healthy weight	1,186 (63%)	337 (65%)	295 (63%)	200 (65%)	171 (61%)	126 (58%)	57 (60%)
Obese	260 (14%)	68 (13%)	69 (15%)	44 (14%)	38 (14%)	25 (11%)	16 (17%)
Overweight	197 (10%)	57 (11%)	40 (8.5%)	37 (12%)	25 (9.0%)	30 (14%)	8 (8.4%)
Underweight	245 (13%)	53 (10%)	67 (14%)	28 (9.1%)	45 (16%)	38 (17%)	14 (15%)
Unknown	1,788	520	140	231	763	23	111
Height_Categories
Normal	3,297 (90%)	931 (90%)	518 (85%)	464 (86%)	989 (95%)	210 (87%)	185 (90%)
Short	263 (7.2%)	68 (6.6%)	67 (11%)	51 (9.4%)	37 (3.6%)	25 (10%)	15 (7.3%)
Tall	116 (3.2%)	36 (3.5%)	26 (4.3%)	25 (4.6%)	16 (1.5%)	7 (2.9%)	6 (2.9%)
Weight_Categories
Increased Weight	295 (8.0%)	80 (7.7%)	78 (13%)	51 (9.4%)	39 (3.7%)	28 (12%)	19 (9.2%)
Low weight	246 (6.7%)	46 (4.4%)	66 (11%)	39 (7.2%)	42 (4.0%)	37 (15%)	16 (7.8%)
Normal	3,135 (85%)	909 (88%)	467 (76%)	450 (83%)	961 (92%)	177 (73%)	171 (83%)
¹ Median (IQR); n (%)

after 1 year

BMI %>% 
  filter(!is.na(bmi)) %>%
  filter(months>=12) %>% 
  arrange(months) %>% 
  group_by(mrn) %>% 
  slice(1) %>% 
  ungroup() %>% 
  dplyr::select(category, BMI_Categories, Height_Categories, Weight_Categories ) %>% 
  tbl_summary(by=category) %>% add_overall()

Characteristic	Overall, N = 2,916¹	Leukemia, N = 891¹	Lymphoma, N = 527¹	CNS, N = 372¹	Solid, N = 778¹	Bone, N = 188¹	STS, N = 160¹
BMI_Categories
Healthy weight	961 (62%)	269 (58%)	255 (62%)	164 (68%)	127 (60%)	91 (58%)	55 (77%)
Obese	246 (16%)	93 (20%)	71 (17%)	24 (10%)	28 (13%)	24 (15%)	6 (8.5%)
Overweight	210 (14%)	74 (16%)	55 (13%)	30 (13%)	27 (13%)	20 (13%)	4 (5.6%)
Underweight	137 (8.8%)	30 (6.4%)	28 (6.8%)	22 (9.2%)	28 (13%)	23 (15%)	6 (8.5%)
Unknown	1,362	425	118	132	568	30	89
Height_Categories
Normal	2,607 (89%)	809 (91%)	447 (85%)	308 (83%)	728 (94%)	167 (89%)	148 (93%)
Short	248 (8.5%)	64 (7.2%)	62 (12%)	53 (14%)	39 (5.0%)	20 (11%)	10 (6.3%)
Tall	61 (2.1%)	18 (2.0%)	18 (3.4%)	11 (3.0%)	11 (1.4%)	1 (0.5%)	2 (1.3%)
Weight_Categories
Increased Weight	264 (9.1%)	96 (11%)	82 (16%)	28 (7.5%)	30 (3.9%)	23 (12%)	5 (3.1%)
Low weight	175 (6.0%)	39 (4.4%)	29 (5.5%)	34 (9.1%)	37 (4.8%)	26 (14%)	10 (6.3%)
Normal	2,477 (85%)	756 (85%)	416 (79%)	310 (83%)	711 (91%)	139 (74%)	145 (91%)
¹ n (%)

after 5 year

BMI%>% 
  filter(!is.na(bmi)) %>%
  filter(months>=60) %>% 
  arrange(months) %>% 
  group_by(mrn) %>% 
  slice(1) %>% 
  ungroup() %>% 
  dplyr::select(category, BMI_Categories, Height_Categories, Weight_Categories ) %>% 
  tbl_summary(by=category) %>% add_overall()

Characteristic	Overall, N = 1,096¹	Leukemia, N = 390¹	Lymphoma, N = 185¹	CNS, N = 173¹	Solid, N = 250¹	Bone, N = 58¹	STS, N = 40¹
BMI_Categories
Healthy weight	491 (61%)	195 (60%)	62 (60%)	96 (66%)	99 (62%)	22 (63%)	17 (55%)
Obese	158 (20%)	72 (22%)	21 (20%)	25 (17%)	26 (16%)	8 (23%)	6 (19%)
Overweight	103 (13%)	48 (15%)	12 (12%)	14 (9.6%)	20 (13%)	2 (5.7%)	7 (23%)
Underweight	48 (6.0%)	10 (3.1%)	8 (7.8%)	11 (7.5%)	15 (9.4%)	3 (8.6%)	1 (3.2%)
Unknown	296	65	82	27	90	23	9
Height_Categories
Normal	908 (83%)	334 (86%)	163 (88%)	122 (71%)	206 (82%)	51 (88%)	32 (80%)
Short	141 (13%)	37 (9.5%)	17 (9.2%)	45 (26%)	29 (12%)	7 (12%)	6 (15%)
Tall	47 (4.3%)	19 (4.9%)	5 (2.7%)	6 (3.5%)	15 (6.0%)	0 (0%)	2 (5.0%)
Weight_Categories
Increased Weight	156 (14%)	71 (18%)	21 (11%)	22 (13%)	29 (12%)	7 (12%)	6 (15%)
Low weight	79 (7.2%)	15 (3.8%)	12 (6.5%)	27 (16%)	17 (6.8%)	5 (8.6%)	3 (7.5%)
Normal	861 (79%)	304 (78%)	152 (82%)	124 (72%)	204 (82%)	46 (79%)	31 (78%)
¹ n (%)

BMI progress over time

I made new figures … look after these 2 bar charts

BMI %>% 
  mutate(BMI_Categories=factor(BMI_Categories, levels=c("Underweight", "Healthy weight", "Overweight", "Obese"))) %>% 
  mutate(year=cut(months, breaks=seq(0,120,12), labels=1:10)) %>% 
  filter(!is.na(BMI_Categories), !is.na(year)) %>% 
  ggplot(aes(x=year, fill=BMI_Categories))+
  geom_bar(position="fill")+
  # scale_x_continuous(limits = c(0, 180), breaks = seq(0, 180, 12), labels=seq(0,15,1))+
  theme_classic()+
  labs(x="Years after diagnosis", y="Proportion", fill="Weight categories")

BMI %>% 
  mutate(BMI_Categories=factor(BMI_Categories, levels=c("Underweight", "Healthy weight", "Overweight", "Obese"))) %>% 
  mutate(year=cut(months, breaks=seq(0,120,12), labels=1:10)) %>% 
  filter(!is.na(BMI_Categories), !is.na(year)) %>% 
  ggplot(aes(x=year, fill=BMI_Categories))+
  geom_bar(position="fill")+
  # scale_x_continuous(limits = c(0, 180), breaks = seq(0, 180, 12), labels=seq(0,15,1))+
  theme_classic()+
  labs(x="Years after diagnosis", y="Proportion", fill="Weight categories")+
  facet_wrap(~category)

BMI progress over time shown using regression lines

BMI %>% 
  mutate(BMI_Categories=factor(BMI_Categories, levels=c("Underweight", "Healthy weight", "Overweight", "Obese"))) %>% 
  mutate(year=cut(months, breaks=seq(0,120,12), labels=1:10)) %>% 
  filter(!is.na(BMI_Categories), !is.na(year)) %>%
  group_by(year, mrn) %>% 
  summarise(bfs=mean(bfs)) %>% 
  mutate(BMI_Categories = case_when(
      bfs < 0.05 ~ "Underweight",
      bfs >= 0.05 & bfs < 0.85 ~ "Healthy weight",
      bfs >= 0.85 & bfs < 0.95 ~ "Overweight",
      bfs >= 0.95 ~ "Obese",
      TRUE ~ NA_character_)) %>% 
  mutate(BMI_Categories=factor(BMI_Categories, levels=c("Healthy weight", "Obese", "Overweight", "Underweight"))) %>% 
  group_by(year) %>% 
  count(BMI_Categories) %>% 
  mutate(total=sum(n), prcnt=n/total) %>% 
  ungroup() %>% 
  ggplot(aes(x=year, color=BMI_Categories, group=BMI_Categories, y=prcnt))+
  geom_point(show.legend = FALSE)+
  stat_smooth(method = "lm", formula='y ~ x', se = T)+
  stat_cor(aes(color = BMI_Categories), method = "spearman", label.x=3.5, label.y=c(0.42,0.38,0.335 ,0.29), )+
  # scale_x_continuous(limits = c(0, 180), breaks = seq(0, 180, 12), labels=seq(0,15,1))+
  theme_classic()+
  labs(x="Years after diagnosis", y="Proportion", fill="Weight categories")+
  theme(legend.position = c(0.2,0.5))

figure below is just to get the R and p values

BMI %>% 
  mutate(BMI_Categories=factor(BMI_Categories, levels=c("Underweight", "Healthy weight", "Overweight", "Obese"))) %>% 
  mutate(year=cut(months, breaks=seq(0,120,12), labels=1:10)) %>% 
  filter(!is.na(BMI_Categories), !is.na(year)) %>%
  group_by(category, year, mrn) %>% 
  summarise(bfs=mean(bfs)) %>% 
  mutate(BMI_Categories = case_when(
      bfs < 0.05 ~ "Underweight",
      bfs >= 0.05 & bfs < 0.85 ~ "Healthy weight",
      bfs >= 0.85 & bfs < 0.95 ~ "Overweight",
      bfs >= 0.95 ~ "Obese",
      TRUE ~ NA_character_)) %>% 
  mutate(BMI_Categories=factor(BMI_Categories, levels=c("Healthy weight", "Obese", "Overweight", "Underweight"))) %>% 
  group_by(category, year) %>% 
  count(BMI_Categories) %>% 
  mutate(total=sum(n), prcnt=n/total) %>% 
  ungroup() %>% 
  ggplot(aes(x=year, color=BMI_Categories, group=BMI_Categories, y=prcnt))+
  geom_point(show.legend = FALSE)+
  stat_smooth(method = "lm", formula='y ~ x', se = T)+
  stat_cor(aes(color = BMI_Categories), method = "spearman", label.x=3.5 )+
  # scale_x_continuous(limits = c(0, 180), breaks = seq(0, 180, 12), labels=seq(0,15,1))+
  theme_classic()+
  labs(x="Years after diagnosis", y="Proportion", fill="Weight categories")+
  theme(legend.position = "none")+
  facet_wrap(~category)

compare first BMI and BMI after 5 years

BMI %>%     filter(!is.na(bmi))%>%      arrange(months) %>%   group_by(mrn) %>% slice(1) %>% mutate(time="Dx") -> BMI_at_diagnosis
BMI %>%     filter(!is.na(bmi))%>%  filter(months>=60) %>%   arrange(months) %>%   group_by(mrn) %>% slice(1)  %>% 
  mutate(time="5-years")  -> BMI_at_5
BMI %>%     filter(!is.na(bmi))%>%  filter(months>=120) %>%   arrange(months) %>%   group_by(mrn) %>% slice(1)  %>% mutate(time="10-years") -> BMI_at_10

combined_df<-dplyr::bind_rows(BMI_at_diagnosis,BMI_at_5,BMI_at_10)


combined_df %>%
  filter(bmi<80) %>% 
  dplyr::select(time, bmi, category, mrn) %>% 
  mutate(time=factor(time, levels=c("Dx", "5-years", "10-years"))) %>% 
  ggplot(aes(x = time, y = bmi, group = mrn)) +
  geom_line(alpha=0.5, color = "grey") +
  geom_point(size=0.1)+
  theme_classic()+
  facet_wrap(~category)+
  labs(y="BMI", x="Timing of Readings")

  theme(strip.background = element_blank())

## List of 1
##  $ strip.background: list()
##   ..- attr(*, "class")= chr [1:2] "element_blank" "element"
##  - attr(*, "class")= chr [1:2] "theme" "gg"
##  - attr(*, "complete")= logi FALSE
##  - attr(*, "validate")= logi TRUE

paired t test

compare Dx and 5 years

To compare BMI at diagnosis and at 5 years for 1096 patients with both readings available, a paired t test showed statistically significant difference in the mean with mean difference of 3.3 (95% CI, 1.59-4.97, p<0.001)

a<-  combined_df %>% filter(time!="10-years") %>% filter(duplicated(mrn) | duplicated(mrn, fromLast = TRUE))

  
t.test(bmi~time, data=a  , paired=T)

## 
##  Paired t-test
## 
## data:  bmi by time
## t = 3.8192, df = 1095, p-value = 0.0001414
## alternative hypothesis: true mean difference is not equal to 0
## 95 percent confidence interval:
##  1.596831 4.971208
## sample estimates:
## mean difference 
##        3.284019

compare 5 and 10 years

To compare BMI at diagnosis and at 5 years for 179 patients with both readings available, a paired t test showed statistically significant difference in the mean with mean difference of 3.2 (95% CI, 2.63-3.76, p<0.001)

a<-  combined_df %>% filter(time!="Dx") %>% filter(duplicated(mrn) | duplicated(mrn, fromLast = TRUE))

  
t.test(bmi~time, data=a  , paired=T)

## 
##  Paired t-test
## 
## data:  bmi by time
## t = 11.108, df = 178, p-value < 2.2e-16
## alternative hypothesis: true mean difference is not equal to 0
## 95 percent confidence interval:
##  2.626559 3.761433
## sample estimates:
## mean difference 
##        3.193996

was there any correlation over time for BMI

Do not include these 2 curves

Spearman correlation

sp <- ggscatter(BMI %>% filter(bmi<80, months<120), x = "months", y = "bmi",
   palette = "gray", alpha=0.01, size=1,
   add = "reg.line", conf.int = TRUE)
sp + stat_cor(method = "spearman", label.x = 80)

sp <- ggscatter(BMI %>% filter(bmi<80, months<120), x = "months", y = "bmi",
   color = "category", palette = "jco", alpha=0.03, size=1,
   add = "reg.line", conf.int = TRUE)
sp + stat_cor(aes(color = category), method = "spearman", label.x = 80)

predictors for obesity at 5 years using logistic regression

This logistic regression analysis aimed to identify predictors of obesity five years following the diagnosis of childhood cancer, considering factors such as gender, age at diagnosis, cancer category, and initial BMI categories. The findings indicated that age at diagnosis and initial BMI category were significant predictors of later obesity. Specifically, younger children at the time of cancer diagnosis were more likely to become obese, with a notable decrease in the odds of obesity for each additional year of age at diagnosis (multivariable OR = 0.78, p < 0.001). Furthermore, children categorized as overweight or obese at diagnosis had significantly higher odds of being obese five years later (multivariable ORs = 4.34 and 8.91, respectively, p < 0.001 for both), highlighting the strong predictive value of initial BMI status. In contrast, gender and specific cancer categories did not maintain significant associations with obesity in the multivariable analysis, indicating that these factors may be less critical in predicting long-term obesity outcomes in this population.

BMI_at_5 %>% filter(bfs>0.95) %>%  pull(mrn)->ObeseMRN
BMI_at_diagnosis$Obesity<-ifelse(BMI_at_diagnosis$mrn %in% ObeseMRN, "YES", "NO")
BMI_at_diagnosis$Obesity<- as.factor(BMI_at_diagnosis$Obesity)
BMI_at_diagnosis$category<- forcats::fct_relevel(BMI_at_diagnosis$category, "Solid")
dependent = "Obesity"
explanatory=c("gender", "age_at_diagnosis", "category", "BMI_Categories")
explanatory_multi=c("age_at_diagnosis", "category", "BMI_Categories")
BMI_at_diagnosis$gender=factor(BMI_at_diagnosis$gender, levels=c("MALE", "FEMALE"))
BMI_at_diagnosis %>% finalfit(dependent, explanatory, explanatory_multi) %>% gt()

Dependent: Obesity		NO	YES	OR (univariable)	OR (multivariable)
gender	MALE	1991 (95.4)	95 (4.6)	-	-
	FEMALE	1527 (96.0)	63 (4.0)	0.86 (0.62-1.19, p=0.381)	-
age_at_diagnosis	Mean (SD)	7.1 (5.5)	5.1 (3.4)	0.93 (0.90-0.96, p<0.001)	0.78 (0.72-0.84, p<0.001)
category	Solid	1015 (97.4)	27 (2.6)	-	-
	Leukemia	963 (93.0)	72 (7.0)	2.81 (1.81-4.48, p<0.001)	1.48 (0.68-3.50, p=0.340)
	Lymphoma	590 (96.6)	21 (3.4)	1.34 (0.74-2.38, p=0.324)	1.22 (0.51-3.11, p=0.661)
	CNS	515 (95.4)	25 (4.6)	1.82 (1.04-3.18, p=0.033)	0.78 (0.30-2.05, p=0.607)
	Bone	234 (96.7)	8 (3.3)	1.29 (0.54-2.74, p=0.540)	1.03 (0.30-3.21, p=0.966)
	STS	201 (97.6)	5 (2.4)	0.94 (0.31-2.26, p=0.892)	1.13 (0.23-4.19, p=0.868)
BMI_Categories	Healthy weight	1162 (98.0)	24 (2.0)	-	-
	Obese	231 (88.8)	29 (11.2)	6.08 (3.48-10.71, p<0.001)	8.91 (4.95-16.24, p<0.001)
	Overweight	180 (91.4)	17 (8.6)	4.57 (2.37-8.63, p<0.001)	4.34 (2.21-8.36, p<0.001)
	Underweight	245 (100.0)		0.00 (0.00-41526.24, p=0.982)	0.00 (0.00-13494.57, p=0.981)

predictors for short stature at 5 years using logistic regression

BMI_at_5 %>% filter(hfs<0.05) %>% pull(mrn)->ShortMRN
BMI_at_diagnosis$short<-ifelse(BMI_at_diagnosis$mrn %in% ShortMRN, "YES", "NO")
BMI_at_diagnosis$short<- as.factor(BMI_at_diagnosis$short)
BMI_at_diagnosis$category<- forcats::fct_relevel(BMI_at_diagnosis$category, "Solid")
BMI_at_diagnosis$gender=factor(BMI_at_diagnosis$gender, levels=c("MALE", "FEMALE"))
dependent = "short"
explanatory=c("gender", "age_at_diagnosis", "category", "Height_Categories")
BMI_at_diagnosis %>% finalfit(dependent, explanatory) %>% gt()

Dependent: short		NO	YES	OR (univariable)	OR (multivariable)
gender	MALE	2021 (96.9)	65 (3.1)	-	-
	FEMALE	1514 (95.2)	76 (4.8)	1.56 (1.11-2.19, p=0.010)	1.55 (1.10-2.21, p=0.013)
age_at_diagnosis	Mean (SD)	7.1 (5.5)	6.7 (3.3)	0.99 (0.96-1.02, p=0.415)	0.92 (0.89-0.96, p<0.001)
category	Solid	1013 (97.2)	29 (2.8)	-	-
	Leukemia	996 (96.2)	39 (3.8)	1.37 (0.84-2.25, p=0.209)	1.48 (0.89-2.46, p=0.130)
	Lymphoma	594 (97.2)	17 (2.8)	1.00 (0.53-1.81, p=0.999)	1.14 (0.58-2.21, p=0.694)
	CNS	498 (92.2)	42 (7.8)	2.95 (1.82-4.83, p<0.001)	3.05 (1.84-5.12, p<0.001)
	Bone	235 (97.1)	7 (2.9)	1.04 (0.42-2.27, p=0.926)	1.26 (0.48-2.98, p=0.612)
	STS	199 (96.6)	7 (3.4)	1.23 (0.49-2.69, p=0.630)	1.35 (0.53-3.02, p=0.495)
Height_Categories	Normal	3201 (97.1)	96 (2.9)	-	-
	Short	218 (82.9)	45 (17.1)	6.88 (4.67-10.01, p<0.001)	10.01 (6.34-15.83, p<0.001)
	Tall	116 (100.0)		0.00 (0.00-0.00, p=0.969)	0.00 (0.00-0.00, p=0.969)

growth parameter after 1 year of diagnosis per age

lower and upper values at diagnosis

MALES

# percentiles 1-5 year after diagnosis
BMI %>%
  filter(age>=0, gender=="MALE") %>% 
  arrange(months) %>% 
  group_by(mrn) %>% slice(1) %>% 
  mutate(ageCategory=cut(age, breaks=c(0:30), labels=seq(0,29,1), right=F)) %>% 
  mutate(yearsAfterDiagnosis=months/12) %>% 
  mutate(yearsCategory=cut(yearsAfterDiagnosis, breaks=c(0:15), labels=c(1:15))) %>% 
  group_by(ageCategory) %>% 
  summarise(Patients=n_distinct(mrn), readings=n(),
            lowerWeightPercentile=quantile(weight, 0.10,na.rm=T), upperWeightPercentile=quantile(weight, 0.90,na.rm=T),
            lowerHeightPercentile=quantile(height, 0.10,na.rm=T), upperHeightPercentile=quantile(height, 0.90,na.rm=T),
            lowerBMIPercentile=quantile(bmi, 0.10,na.rm=T), upperBMIPercentile=quantile(bmi, 0.90,na.rm=T))

ageCategory	Patients	readings	lowerWeightPercentile	upperWeightPercentile	lowerHeightPercentile	upperHeightPercentile	lowerBMIPercentile	upperBMIPercentile
0	144	144	5.6	11	55.7	78.3	14.7	20.7
1	170	170	9	13.7	75.1	88	14.6	19.6
2	173	173	11.7	16.7	83	99	14.7	18.7
3	163	163	12.9	19.2	92	108	14	18.3
4	170	170	15	22	99	113	14.1	17.9
5	154	154	16	26.9	104	120	13.7	18.8
6	116	116	17.7	27.5	110	125	13.6	18.3
7	96	96	19	32.2	114	131	13.6	19.5
8	80	80	21.4	32.3	121	138	13.6	19.7
9	97	97	24.1	44.1	125	143	14.2	22.6
10	83	83	26	52.3	128	147	14.3	24.1
11	86	86	28.7	59.4	135	157	14.9	27.3
12	83	83	30.9	63.7	141	164	15.1	25.7
13	88	88	33.5	76	143	168	15.1	28.6
14	72	72	38.4	81.8	152	178	16	28.1
15	109	109	43.8	86	158	178	16.8	30.6
16	77	77	46.3	93.6	161	178	17.5	30.9
17	109	109	51.6	96.2	163	179	18.3	31.4
18	21	21	51.9	91.5	163	177	18.8	28.9
19	7	7	51.4	74	169	181	17.5	23.4
20	7	7	49.2	86.3	169	179	20.6	27.8
21	2	2	63	63	170	170	21.8	21.8
22	3	3	79	87.8	176	177	25.8	28.4
23	1	1	59	59	173	173	19.7	19.7

FEMALES

# percentiles 1-5 year after diagnosis
BMI %>%
  filter(age>=0, gender=="FEMALE") %>% 
  arrange(months) %>% 
  group_by(mrn) %>% slice(1) %>% 
  mutate(ageCategory=cut(age, breaks=c(0:30), labels=seq(0,29,1), right=F)) %>% 
  mutate(yearsAfterDiagnosis=months/12) %>% 
  mutate(yearsCategory=cut(yearsAfterDiagnosis, breaks=c(0:15), labels=c(1:15))) %>% 
  group_by(ageCategory) %>% 
  summarise(Patients=n_distinct(mrn), readings=n(),
            lowerWeightPercentile=quantile(weight, 0.10,na.rm=T), upperWeightPercentile=quantile(weight, 0.90,na.rm=T),
            lowerHeightPercentile=quantile(height, 0.10,na.rm=T), upperHeightPercentile=quantile(height, 0.90,na.rm=T),
            lowerBMIPercentile=quantile(bmi, 0.10,na.rm=T), upperBMIPercentile=quantile(bmi, 0.90,na.rm=T))

ageCategory	Patients	readings	lowerWeightPercentile	upperWeightPercentile	lowerHeightPercentile	upperHeightPercentile	lowerBMIPercentile	upperBMIPercentile
0	120	120	4.39	10	50	76	14.2	19.8
1	134	134	8.91	13.4	72.9	87.1	15	19.3
2	148	148	11	15.9	84	97.5	14.2	18.1
3	126	126	12.6	18.1	91	104	13.4	18.6
4	132	132	13.5	20.1	98	109	13.6	18.2
5	96	96	15.1	23.7	103	118	13.1	18
6	75	75	16.8	26.9	106	124	14.1	18.5
7	84	84	18.9	31.6	116	129	13.4	20.1
8	53	53	20.8	34.1	119	138	13.6	19
9	64	64	24	44.8	125	143	13.6	23.4
10	61	61	25	50	129	150	14.5	24.6
11	58	58	26.9	58.9	131	155	14.3	24.5
12	72	72	33.7	60.1	143	162	15	25.5
13	70	70	36.9	59.7	146	163	15.9	24.3
14	61	61	39.5	74.2	146	164	17.6	28.3
15	68	68	44.6	82.5	151	167	17.8	30
16	71	71	41.8	70	150	167	17.8	27.7
17	78	78	42.3	81.8	152	168	17.5	29.2
18	18	18	47.3	88.7	156	171	18.6	33.2
19	7	7	47.8	74	157	166	19.2	27.4
20	4	4	55.4	112	164	169	22.8	41.9
21	6	6	50	76.5	157	166	19.7	27.7
22	1	1	49	49	161	161	18.9	18.9
24	2	2	45.9	61.1	163	167	16.6	22.9
26	1	1	58	58	153	153	24.8	24.8

lower and upper values at1-5 years afte diagnosis

MALES

# percentiles 1-5 year after diagnosis
BMI %>%
  filter(age>=0, months>=12, months<48, gender=="MALE") %>% 
  mutate(ageCategory=cut(age, breaks=c(0:30), labels=seq(0,29,1), right=F)) %>% 
  mutate(yearsAfterDiagnosis=months/12) %>% 
  mutate(yearsCategory=cut(yearsAfterDiagnosis, breaks=c(0:15), labels=c(1:15))) %>% 
  group_by(ageCategory) %>% 
  summarise(Patients=n_distinct(mrn), readings=n(),
            lowerWeightPercentile=quantile(weight, 0.10,na.rm=T), upperWeightPercentile=quantile(weight, 0.90,na.rm=T),
            lowerHeightPercentile=quantile(height, 0.10,na.rm=T), upperHeightPercentile=quantile(height, 0.90,na.rm=T),
            lowerBMIPercentile=quantile(bmi, 0.10,na.rm=T), upperBMIPercentile=quantile(bmi, 0.90,na.rm=T))

ageCategory	Patients	readings	lowerWeightPercentile	upperWeightPercentile	lowerHeightPercentile	upperHeightPercentile	lowerBMIPercentile	upperBMIPercentile
1	117	1515	9	14.4	76	91	15.1	20
2	208	2761	11.7	18.7	84	100	15.3	20.2
3	303	3517	13.5	20.8	91.5	107	15.1	19.5
4	305	3874	15.5	23.8	99	114	14.6	19.3
5	304	3810	16.9	26.5	104	121	14.5	19.4
6	266	2760	18.2	30.2	110	128	14.4	19.2
7	228	2690	20.1	35.5	115	132	14.6	21
8	219	2549	22	41.1	118	138	14.8	23.2
9	200	2144	23	46	122	143	14.4	24.1
10	171	1581	26	48	128	148	14.9	23.6
11	161	1430	27	56.3	132	153	15.2	25
12	155	1945	28.9	69.5	136	158	15.9	28.5
13	154	2108	34.3	80	143	167	16.4	30.5
14	160	1904	40.6	96	151	173	16.8	33.2
15	170	1739	42	91.8	156	175	17	30.7
16	184	2110	46.6	90	159	176	17.6	30.8
17	183	2078	48.6	97	160	177	17.4	33.2
18	188	2457	52.5	98.3	161	179	18.3	31.9
19	127	1777	55	100	163	177	18.7	33.4
20	75	783	51	97.2	165	178	18.7	31.6
21	27	179	57	81.8	164	180	19.4	28.1

FEMALES

# percentiles 1-5 year after diagnosis
BMI %>%
  filter(age>=0, months>=12, months<48, gender=="FEMALE") %>% 
  mutate(ageCategory=cut(age, breaks=c(0:30), labels=seq(0,29,1), right=F)) %>% 
  mutate(yearsAfterDiagnosis=months/12) %>% 
  mutate(yearsCategory=cut(yearsAfterDiagnosis, breaks=c(0:15), labels=c(1:15))) %>% 
  group_by(ageCategory) %>% 
  summarise(Patients=n_distinct(mrn), readings=n(),
            lowerWeightPercentile=quantile(weight, 0.10,na.rm=T), upperWeightPercentile=quantile(weight, 0.90,na.rm=T),
            lowerHeightPercentile=quantile(height, 0.10,na.rm=T), upperHeightPercentile=quantile(height, 0.90,na.rm=T),
            lowerBMIPercentile=quantile(bmi, 0.10,na.rm=T), upperBMIPercentile=quantile(bmi, 0.90,na.rm=T))

ageCategory	Patients	readings	lowerWeightPercentile	upperWeightPercentile	lowerHeightPercentile	upperHeightPercentile	lowerBMIPercentile	upperBMIPercentile
1	102	936	9.5	14.9	76	91	15.2	19.1
2	175	1787	11.6	17.8	85	99.1	15.1	19.4
3	247	2709	13.3	19.4	91.1	105	14.7	19.1
4	229	2664	14.9	23	97	113	14.4	19.1
5	240	3146	16.5	25.2	104	118	14.1	19.2
6	204	2374	18	28.8	108	124	14.2	19.8
7	171	1923	20.1	35	114	130	14.4	21.4
8	141	1620	20.8	38.7	119	136	14	22.2
9	119	1282	24	41.1	125	140	15	21.8
10	120	1296	27	46	128	145	15.1	24.3
11	119	1199	31.5	56.3	132	152	15.7	28
12	117	1197	27	60.5	133	159	15.3	26
13	128	1157	36.8	66	145	164	16.4	26.5
14	131	1169	38.7	65	147	163	16.9	26.4
15	134	1461	40.5	69	149	165	17.7	27.2
16	125	1895	39.6	71.5	149	165	16.8	28.3
17	141	1331	46	81.3	151	166	18.7	32.3
18	151	1378	45.6	82	153	166	17.9	32.4
19	105	1034	48	85.6	153	167	18.6	32.9
20	60	513	47	121	154	168	18.2	35.5
21	21	85	48.1	81.5	153	166	20	31.5

lower and upper values after 5 years of diagnosis

# percentiles 5 and more years after diagnoss
BMI %>%
  filter(age>=5, age<20, months>=60) %>% 
  mutate(ageCategory=cut(age, breaks=c(0:30), labels=seq(0,29,1), right=F)) %>% 
  mutate(yearsAfterDiagnosis=months/12) %>% 
  mutate(yearsCategory=cut(yearsAfterDiagnosis, breaks=c(0:15), labels=c(1:15))) %>% 
  group_by(ageCategory) %>% 
  summarise(Patients=n_distinct(mrn), readings=n(),
            lowerWeightPercentile=quantile(weight, 0.10,na.rm=T), upperWeightPercentile=quantile(weight, 0.90,na.rm=T),
            lowerHeightPercentile=quantile(height, 0.10,na.rm=T), upperHeightPercentile=quantile(height, 0.90,na.rm=T),
            lowerBMIPercentile=quantile(bmi, 0.10,na.rm=T), upperBMIPercentile=quantile(bmi, 0.90,na.rm=T))

ageCategory	Patients	readings	lowerWeightPercentile	upperWeightPercentile	lowerHeightPercentile	upperHeightPercentile	lowerBMIPercentile	upperBMIPercentile
5	93	627	18	24.6	106	122	14.4	19.2
6	154	783	19.2	31.9	111	125	14.3	21.6
7	230	1241	20	41	115	132	14.1	25.5
8	270	1272	22	44.6	118	138	14.7	24.7
9	271	1007	24	44.9	125	142	15.1	23.9
10	277	1251	25.7	54.3	127	150	14.8	26.8
11	257	1321	29.3	71.6	131	154	15.5	29
12	248	1273	31.3	64	136	157	16	28.7
13	215	945	34	70.4	140	162	16.6	28.3
14	184	637	38	74.5	138	166	17.1	29.8
15	155	819	43	77.8	147	169	17.6	30.4
16	143	715	44.5	85.4	149	175	18.2	32
17	147	614	48	109	148	176	18.4	32.8
18	134	545	50.9	89.5	155	178	19.4	33.7
19	136	640	40.1	84	155	177	16.4	29.2

Growth Chart

Cox regression for prediction of mortality using BMI

explanatory=c("gender", "age_at_diagnosis", "category", "BMI_Categories", "Weight_Categories", "Height_Categories")
explanatory_multi=c("age_at_diagnosis", "category")
dependent = "Surv(OS, dead)"

BMI_at_diagnosis %>% finalfit(dependent, explanatory, explanatory_multi) %>% gt()

Dependent: Surv(OS, dead)		all	HR (univariable)	HR (multivariable)
gender	MALE	2086 (56.7)	-	-
	FEMALE	1590 (43.3)	0.99 (0.85-1.15, p=0.889)	-
age_at_diagnosis	Mean (SD)	7.0 (5.4)	1.02 (1.00-1.03, p=0.009)	1.02 (1.01-1.04, p=0.002)
category	Solid	1042 (28.3)	-	-
	Leukemia	1035 (28.2)	1.11 (0.90-1.37, p=0.325)	1.06 (0.85-1.31, p=0.613)
	Lymphoma	611 (16.6)	0.40 (0.29-0.57, p<0.001)	0.35 (0.24-0.49, p<0.001)
	CNS	540 (14.7)	2.01 (1.62-2.50, p<0.001)	1.90 (1.52-2.37, p<0.001)
	Bone	242 (6.6)	2.34 (1.80-3.05, p<0.001)	1.96 (1.48-2.61, p<0.001)
	STS	206 (5.6)	2.26 (1.70-3.01, p<0.001)	2.11 (1.58-2.81, p<0.001)
BMI_Categories	Healthy weight	1186 (62.8)	-	-
	Obese	260 (13.8)	0.82 (0.59-1.14, p=0.234)	-
	Overweight	197 (10.4)	1.02 (0.73-1.44, p=0.888)	-
	Underweight	245 (13.0)	1.20 (0.90-1.61, p=0.215)	-
Weight_Categories	Increased Weight	295 (8.0)	-	-
	Low weight	246 (6.7)	1.26 (0.87-1.83, p=0.220)	-
	Normal	3135 (85.3)	0.99 (0.75-1.31, p=0.946)	-
Height_Categories	Normal	3297 (89.7)	-	-
	Short	263 (7.2)	1.22 (0.93-1.59, p=0.149)	-
	Tall	116 (3.2)	1.04 (0.68-1.59, p=0.859)	-

vitals_feb24

2024-02-26

Methods

Loading data

Donut chart showing disribution of patients

Table showing charachterstics of patients

count of patients per year of diagnosis

median number of readings (height or weight) recorded per patient with 25 and 75% quantiles

Growth paramaters at diagnosis

growth parameters after 1 year

First reading after 5 years of diagnosis

table showing number of patients with abnormal values at diagnosis

after 1 year

after 5 year

BMI progress over time

BMI progress over time shown using regression lines

figure below is just to get the R and p values

compare first BMI and BMI after 5 years

paired t test

compare Dx and 5 years

compare 5 and 10 years

was there any correlation over time for BMI

Do not include these 2 curves

predictors for obesity at 5 years using logistic regression

predictors for short stature at 5 years using logistic regression

growth parameter after 1 year of diagnosis per age

lower and upper values at diagnosis

MALES

FEMALES

lower and upper values at1-5 years afte diagnosis

MALES

FEMALES

lower and upper values after 5 years of diagnosis

Growth Chart

Cox regression for prediction of mortality using BMI