Metabolic Syndrome Prevalence across time using NHANES Data

Biostatistics NHANES approach R ggplot2

How to correctly approach NHANES data to make estimates that are representative of the population
Define Metabolic Syndrome based on ATP III
Look at the MetS prevalence over time

Hai Nguyen
June 08, 2021

I need to use these packages:

library(tidyverse)
library(foreign) 
library(survey) 
options(survey.lonely.psu='adjust')
library(kableExtra)
#cat("R package versions:\n")
for (p in c("base", "survey","tidyverse")) { 
  cat(p, ": ", as.character(packageVersion(p)), "\n")
}
base :  4.0.4 
survey :  4.0 
tidyverse :  1.3.1

NHANES Cycle 2013/2014

Import data from NHANES cycle 2013/2014

Code for data import was adapted from a NHANES tutorials example.

downloadNHANES <- function(fileprefix){
  print (fileprefix)
  outdf <- data.frame(NULL)
  for (j in 1:length(letters)){
    urlstring <- paste('https://wwwn.cdc.gov/nchs/nhanes/',yrs[j],'/',fileprefix,letters[j],'.XPT', sep='')
    download.file(urlstring, tf <- tempfile(), mode="wb")
    tmpframe <- foreign::read.xport(tf)
    outdf <- bind_rows(outdf, tmpframe)
  }
  return(outdf)
}
# Specify the survey cycles required, with corresponding file suffixes
yrs <- c('2013-2014')
letters <- c('_h')

# Download data for each component
# (1) Demographic (DEMO)
DEMO <- downloadNHANES('DEMO') # 10175
# (2) Blood Pressure & Cholesterol
BPQ <- downloadNHANES('BPQ') # 6464
# (3) Diabetes 
DIQ <- downloadNHANES('DIQ') # 9770
# (4) Blood Pressure
BPX <- downloadNHANES('BPX') # 9813
# (5) Body Measures
BMX <- downloadNHANES('BMX') # 9813
# (6) Cholesterol - High - Density Lipoprotein
HDL <- downloadNHANES('HDL') # 8291
# (7) Cholesterol - Low-Density Lipoproteins (LDL) & Triglycerides (TRIGLY_J)
TRIGLY <- downloadNHANES('TRIGLY') # 3329
# (8) Plasma Fasting Glucose (GLU_J)
GLU <- downloadNHANES('GLU') # 3329
NHANES Data Structure 
# Merging using "piping" 
nhanesdata.merged <- left_join(DEMO, BPQ, by="SEQN") %>% 
  left_join(DIQ, by="SEQN") %>% 
  left_join(BPX, by='SEQN') %>% 
  left_join(BMX, by='SEQN') %>%
  left_join(HDL, by='SEQN') %>%
  left_join(TRIGLY, by='SEQN') %>%
  left_join(GLU, by='SEQN') # 10175 obs

Data Processing

nhanes.vbls <- nhanesdata.merged %>%
  select(SEQN, # Respondent sequence number
         RIDAGEYR, # Age in years at screening
         RIAGENDR, # Gender

         RIDRETH3, # Race/Hispanic origin w/ NH Asian
         DMDMARTL, # Marital status
         DMDEDUC2, # Education level - Adults 20+
         
         # Survey design variables
         SDMVPSU, # Masked variance pseudo-PSU
         SDMVSTRA, # Masked variance pseudo-stratum
         WTMEC2YR, # Full sample 2 year MEC exam weight
         WTINT2YR, # Full sample 2 year interview weight
         
         BPXSY1, # Systolic: Blood pres (1st rdg) mm Hg
         BPXDI1, # Diastolic: Blood pres (1st rdg) mm Hg
         BPXSY2, # Systolic: Blood pres (2nd rdg) mm Hg
         BPXDI2, # Diastolic: Blood pres (2nd rdg) mm Hg
         BPXSY3, # Systolic: Blood pres (3rd rdg) mm Hg
         BPXDI3, # Diastolic: Blood pres (3rd rdg) mm Hg
         BPXSY4, # Systolic: Blood pres (4th rdg) mm Hg
         BPXDI4, # Diastolic: Blood pres (4th rdg) mm Hg
         LBXGLU, # Fasting Glucose (mg/dL)
         LBDHDD, # Direct HDL-Cholesterol (mg/dL)
         LBXTR, # Triglyceride (mg/dL)
         BMXWAIST, # Waist Circumference (cm)
         
         BPQ090D, # Told to take prescription for cholesterol
         BPQ040A, # Taking prescription for hypertension
         DIQ070 # Take diabetic pills to lower blood sugar
         ) %>%
  mutate(# create indicator for overall summary
         one = 1,
         id = SEQN,
         psu = SDMVPSU, strata = SDMVSTRA, 
         persWeight = WTINT2YR, persWeightMEC= WTMEC2YR,
         age = RIDAGEYR,
         # Create age categories for adults aged 18 and over: ages 18-39, 40-59, 60 and over
        ageCat = cut(RIDAGEYR,
                    breaks = c(19, 29, 39, 49, 59, 69, Inf),
                    labels = c('20-29','30-39', '40-49', '50-59', '60-69', '70+')),
         female = factor(if_else(RIAGENDR == 1, 0, 1)), # 1:male ==> 0
         ethnicity = case_when(
           RIDRETH3 %in% c(1,2) ~ 0, # Hispanic
           RIDRETH3 == 3 ~ 1, # Non-Hispanic White
           RIDRETH3 == 4 ~ 2, # Non-Hispanic Black
           RIDRETH3 == 6 ~ 3, # Non-Hispanic Asian
           RIDRETH3 == 7 ~ 4, # Multi-Racial
         is.na(RIDRETH3) ~ NA_real_
         ),
         chol.prsn = case_when(
           BPQ090D == 2 ~ 0, # No
           BPQ090D == 1 ~ 1, # Yes
           BPQ090D %in% c(7,9) ~ NA_real_
         ),
         bp.prsn = case_when(
           BPQ040A == 2 ~ 0, # No
           BPQ040A == 1 ~ 1, # Yes
           BPQ040A %in% c(7,9) ~ NA_real_
         ),
         glu.prsn = case_when(
           DIQ070 == 2 ~ 0, # No
           DIQ070 == 1 ~ 1, # Yes
           DIQ070 %in% c(7,9) ~ NA_real_
         ),
         dia = rowMeans(.[, c("BPXDI1","BPXDI2","BPXDI3","BPXDI4")], na.rm = TRUE),
         sys = rowMeans(.[, c("BPXSY1","BPXSY2","BPXSY3","BPXSY4")], na.rm = TRUE),
         glu = LBXGLU,
         hdl = LBDHDD,
         tri = LBXTR,
         waist = BMXWAIST
        ) %>%
  select(one, id, psu, strata, persWeight, persWeightMEC, age, ageCat, female, ethnicity, sys, dia, glu, hdl, tri, waist, bp.prsn, chol.prsn, glu.prsn) # 10175

Define metabolic syndrome

According to Alberti et al. (2009), the presence of any 3 of 5 risk factors constitutes a diagnosis of metabolic syndrome:

  1. elevated waist circumference (≥88 cm for women and ≥102 cm for men),
  2. elevated triglycerides (≥150 mg/dL) or drug treatment for elevated triglycerides (there was no variable indicating the medication for lower triglycerides),
  3. low HDL cholesterol (<40 mg/dL for men and <50 mg/dL for women) or drug treatment for low HDL cholesterol,
  4. elevated blood pressure (systolic ≥130 mm Hg, or diastolic ≥85 mm Hg, or both) or antihypertensive drug treatment for a history of hypertension, and
  5. elevated fasting glucose (≥100 mg/dL) or drug treatment for elevated glucose.” (Moore et al., 2017; Preventing Chronic Disease).

Besides, I intentionally created the indicators for missing data: - return NA if all indicators are NA for mets.sum: when I summed up all 5 MetS indicators, if all are NA, so sum should be NA (not 0)
- ind.non_missing: return TRUE if none of indicators are none-NA, otherwise FALSE

Finally, mets.ind (MetS indicator) would be binary of 1 and 0 if sum of 5 indicators above greater or equal to 3.

nhanes.MetS <- nhanes.vbls %>%
  mutate(waist.ind = case_when(waist >= 88  & female == 1 ~ 1, #condition 1
                               waist >= 102 & female == 0 ~ 1, #condition 2
                               is.na(waist) | is.na(female) ~ NA_real_, #condition 3
                               TRUE ~ 0), #all other cases
         tri.ind = case_when(tri >= 150 | chol.prsn == 1 ~ 1,
                             is.na(tri) ~ NA_real_,
                             TRUE ~0),
         hdl.ind = case_when((hdl < 40 & female == 0) | chol.prsn == 1 ~ 1,
                             (hdl < 50 & female == 1) | chol.prsn == 1 ~ 1,
                             is.na(hdl) | is.na(female) ~ NA_real_,
                             TRUE ~ 0),
         bp.ind = case_when(sys >= 130 | dia >= 85 | bp.prsn == 1 ~ 1,
                            is.na(sys) & is.na(dia) ~ NA_real_,
                            TRUE ~ 0
                            ),
         glu.ind = case_when(glu >= 100 | glu.prsn == 1 ~ 1,
                             is.na(glu) ~ NA_real_,
                             TRUE ~ 0)
         ) # 10175 obs



nhanes.MetS$mets.sum = rowSums(nhanes.MetS[20:24], na.rm = T) # from waist.ind to glu.ind # 20:24
nhanes.MetS$mets.sum[rowSums(!is.na(nhanes.MetS[20:24])) == 0] <- NA # return NA if all indicators are NA
nhanes.MetS$ind.non_missing <- !rowSums(!is.na(nhanes.MetS[20:24])==0) # create an indicator in which none of criteria is none NA
nhanes.MetS$mets.ind <- ifelse(nhanes.MetS$mets.sum >= 3, 1, 0) # 10175 obs

Here we can get the sense of data for all indicators created:

Define sub-population of interest

nhanes.MetS <- nhanes.MetS %>% 
  mutate(
    #  Define sub-population of interest 
    inAnalysis1 = (age >=20 & !is.na(mets.ind)),
    inAnalysis2 = (age >=20 & ind.non_missing)
    ) # 10175 obs
saveRDS(nhanes.MetS, file = "data/nhanesMetS13_14snd.rds")

Our sample size would be if I subset data on:

MetS proportion in the random sample (unweighted)

With all age groups

nhanes.MetS %>%
  group_by(mets.ind) %>% 
  summarise(n=n()) %>% 
  mutate(prop=round(n/sum(n)*100,1)) %>%
  kbl(col.names = c('Metabolic syndrome', 'Freq', 'Percent')) %>% 
  kable_material(c("striped", "hover"))
Metabolic syndrome Freq Percent
0 6940 68.2
1 2184 21.5
NA 1051 10.3

With all age groups & none missing of MetS indicator

nhanes.MetS %>%
  filter(!is.na(mets.ind)) %>% 
  group_by(mets.ind) %>% 
  summarise(n=n()) %>% 
  mutate(prop=round(n/sum(n)*100,1)) %>%
  kbl(col.names = c('Metabolic syndrome', 'Freq', 'Percent')) %>% 
  kable_material(c("striped", "hover"))
Metabolic syndrome Freq Percent
0 6940 76.1
1 2184 23.9

With age 20+ & none missing of MetS indicator

nhanes.MetS %>% 
  filter(age>=20 & !is.na(mets.ind)) %>% 
  group_by(mets.ind) %>% 
  summarise(n=n()) %>% 
  mutate(prop=round(n/sum(n)*100,1)) %>%
  kbl(col.names = c('Metabolic syndrome', 'Freq', 'Percent')) %>% 
  kable_material(c("striped", "hover"))
Metabolic syndrome Freq Percent
0 3511 62.2
1 2138 37.8

With age 20+ and ind.non_missing is TRUE:

Note: ind.non_missing was created an indicator in which none of criteria is none NA

nhanes.MetS %>% 
  filter(age>=20 & ind.non_missing) %>% 
  group_by(mets.ind) %>% 
  summarise(n=n()) %>% 
  mutate(prop=round(n/sum(n)*100,1)) %>%
  kbl(col.names = c('Metabolic syndrome', 'Freq', 'Percent')) %>% 
  kable_material(c("striped", "hover"))
Metabolic syndrome Freq Percent
0 1273 49.2
1 1314 50.8

MetS prevalence with weighted

Note:  
psu = SDMVPSU, 
strata = SDMVSTRA, 
persWeight = WTINT2YR, 
persWeightMEC= WTMEC2YR
# Define survey design for overall dataset 
NHANES_all <- svydesign(data=nhanes.MetS, id=~psu, strata=~strata, weights=~persWeightMEC, nest=TRUE)
# Create a survey design object for the subset of interest 
# Subsetting the original survey design object ensures we keep the design information about the number of clusters and strata
NHANES <- subset(NHANES_all, inAnalysis1==1)

Proportion and SE, for adults aged 20+

#' Proportion and standard error, for adults aged 20 and over
svyby(~mets.ind, ~one, NHANES, svymean) %>% mutate(mets.ind = round(mets.ind*100, digits=1), se=round(se*100, digits=1))
  one mets.ind  se
1   1       36 0.8

Proportion and SE, for adults age groups

#' Proportion and standard error, for adults age groups
svyby(~mets.ind, ~ageCat, NHANES, svymean) %>% mutate(mets.ind = round(mets.ind*100, digits=1), se=round(se*100, digits=1))
      ageCat mets.ind  se
20-29  20-29      7.0 1.0
30-39  30-39     21.6 1.0
40-49  40-49     30.0 1.3
50-59  50-59     45.7 2.1
60-69  60-69     59.8 2.8
70+      70+     67.2 1.4

Proportion and SE, for adults aged 20+ by race and Hispanic origin

#' Proportion and standard error, for adults aged 20+ by race and Hispanic origin
svyby(~mets.ind, ~ethnicity, NHANES, svymean) %>% mutate(mets.ind = round(mets.ind*100, digits=1), se=round(se*100, digits=1))
  ethnicity mets.ind  se
0         0     30.1 1.1
1         1     39.0 1.1
2         2     34.1 1.6
3         3     23.4 2.8
4         4     28.3 3.9

Cycle 2015/2016

I processed the same procedures as the cycle 2013/2014. Then saved as RDS file. Here I just showed the prevalence after reading the RDS data into R using weights.

nhanes.MetS <- readRDS(file = "data/nhanesMetS15_16.rds")
NHANES_all <- svydesign(data=nhanes.MetS, id=~psu, strata=~strata, weights=~persWeightMEC, nest=TRUE)
NHANES <- subset(NHANES_all, inAnalysis1==1)

Proportion and SE, for adults aged 20+

#' Proportion and standard error, for adults aged 20 and over
svyby(~mets.ind, ~one, NHANES, svymean) %>% mutate(mets.ind = round(mets.ind*100, digits=1), se=round(se*100, digits=1))
  one mets.ind  se
1   1     37.7 1.8

Proportion and SE, for adults age groups

#' Proportion and standard error, for adults age groups
svyby(~mets.ind, ~ageCat, NHANES, svymean) %>% mutate(mets.ind = round(mets.ind*100, digits=1), se=round(se*100, digits=1))
      ageCat mets.ind  se
20-29  20-29      9.2 1.5
30-39  30-39     19.3 1.4
40-49  40-49     33.2 3.2
50-59  50-59     48.1 2.5
60-69  60-69     61.7 2.4
70+      70+     66.7 2.5

Proportion and SE, for adults aged 20+ by race and Hispanic origin

#' Proportion and standard error, for adults aged 20+ by race and Hispanic origin
svyby(~mets.ind, ~ethnicity, NHANES, svymean) %>% mutate(mets.ind = round(mets.ind*100, digits=1), se=round(se*100, digits=1))
  ethnicity mets.ind  se
0         0     34.0 1.8
1         1     40.1 2.4
2         2     34.0 1.2
3         3     23.9 2.8
4         4     43.9 4.1

Cycle 2017/2018

nhanes.MetS <- readRDS(file = "data/nhanesMetS17_18.rds")
NHANES_all <- svydesign(data=nhanes.MetS, id=~psu, strata=~strata, weights=~persWeightMEC, nest=TRUE)
NHANES <- subset(NHANES_all, inAnalysis1==1)

Proportion and SE, for adults aged 20+

#' Proportion and standard error, for adults aged 20 and over
svyby(~mets.ind, ~one, NHANES, svymean) %>% mutate(mets.ind = round(mets.ind*100, digits=1), se=round(se*100, digits=1))
  one mets.ind  se
1   1     38.2 1.3

Proportion and SE, for adults age groups

#' Proportion and standard error, for adults age groups
svyby(~mets.ind, ~ageCat, NHANES, svymean) %>% mutate(mets.ind = round(mets.ind*100, digits=1), se=round(se*100, digits=1))
      ageCat mets.ind  se
20-29  20-29      9.0 1.5
30-39  30-39     18.4 1.6
40-49  40-49     31.1 1.8
50-59  50-59     49.2 3.4
60-69  60-69     60.5 2.8
70+      70+     71.3 1.4

Proportion and SE, for adults aged 20+ by race and Hispanic origin

#' Proportion and standard error, for adults aged 20+ by race and Hispanic origin
svyby(~mets.ind, ~ethnicity, NHANES, svymean) %>% mutate(mets.ind = round(mets.ind*100, digits=1), se=round(se*100, digits=1))
  ethnicity mets.ind  se
0         0     34.5 1.7
1         1     40.3 1.7
2         2     33.5 1.7
3         3     32.1 2.3
4         4     42.5 4.4

Summary

Cycle <- c("2013-2014","2015-2016","2017-2018")
Prevalence <- c(36, 37.7, 38.2)
year.prl <- as.tibble(data.frame(Cycle, Prevalence))

Age.Categories <- c('20-29','30-39','40-49','50-59','60-69','70+')
Prevalence.Cycle.13_14 <- c(7,21.6,30,45.7,59.8,67.2)
Prevalence.Cycle.15_16 <- c(9.2,19.3,33.2,48.1,61.7,66.7)
Prevalence.Cycle.17_18 <- c(9,18.4,31.1,49.2,60.5,71.3)
age.g.p <- as.tibble(data.frame(Age.Categories,Prevalence.Cycle.13_14,Prevalence.Cycle.15_16,Prevalence.Cycle.17_18))
age.g.p.l <- gather(age.g.p, Prevalence, Percent, Prevalence.Cycle.13_14:Prevalence.Cycle.17_18)

Race.Ethnic.Categories <- c('Hispanic','White','Black','Asian','Multi-Racial')
Prevalence.Cycle.13_14 <- c(34.0,40.1,34.0,23.9,43.9)
Prevalence.Cycle.15_16 <- c(34.0,40.1,34.0,23.9,43.9)
Prevalence.Cycle.17_18 <- c(34.5,40.3,33.5,32.1,42.5)
r.e.g.p <- as.tibble(data.frame(Race.Ethnic.Categories,Prevalence.Cycle.13_14,Prevalence.Cycle.15_16,Prevalence.Cycle.17_18))
r.e.g.p.l <- gather(r.e.g.p, Prevalence, Percent, Prevalence.Cycle.13_14:Prevalence.Cycle.17_18)
year.prl %>% ggplot(aes(x = Cycle, y = Prevalence)) + 
  geom_point() +
  geom_line(aes(group=1)) +
  geom_text(label=paste0(Prevalence,"%"), hjust=1.5, vjust=0) +
  ylim(30, 45) +
  theme_bw()

age.g.p.l %>% ggplot(aes(fill=Prevalence, y=Percent, x=Age.Categories)) + 
  geom_bar(position="dodge", stat="identity") +
  theme_bw()

r.e.g.p.l %>% ggplot(aes(fill=Prevalence, y=Percent, x=Race.Ethnic.Categories)) + 
  geom_bar(position="dodge", stat="identity") +
  theme_bw()

References

Alberti, K. G. M. M., Robert H. Eckel, Scott M. Grundy, Paul Z. Zimmet, James I. Cleeman, Karen A. Donato, Jean-Charles Fruchart, W. Philip T. James, Catherine M. Loria, and Jr. Smith Sidney C. 2009. “Harmonizing the Metabolic Syndrome: A Joint Interim Statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity.” Journal Article. Circulation 120 (16): 1640–45. https://doi.org/10.1161/CIRCULATIONAHA.109.192644.
Ford, Earl S., Wayne H. Giles, and William H. Dietz. 2002. “Prevalence of the Metabolic Syndrome Among US Adults: Findings from the Third National Health and Nutrition Examination Survey.” Journal Article. JAMA : The Journal of the American Medical Association 287 (3): 356–59. https://doi.org/10.1001/jama.287.3.356.
Moore, Justin Xavier, Ninad Chaudhary, and Tomi Akinyemiju. 2017. “Metabolic Syndrome Prevalence by Race/Ethnicity and Sex in the United States, National Health and Nutrition Examination Survey, 1988-2012.” Journal Article. Preventing Chronic Disease 14: E24–24. https://doi.org/10.5888/pcd14.160287.
Smiley, Abbas, David King, and Aurelian Bidulescu. 2019. “The Association Between Sleep Duration and Metabolic Syndrome: The NHANES 2013/2014.” Journal Article. Nutrients 11 (11): 2582. https://doi.org/10.3390/nu11112582.

References

Corrections

If you see mistakes or want to suggest changes, please create an issue on the source repository.

Reuse

Text and figures are licensed under Creative Commons Attribution CC BY 4.0. Source code is available at https://github.com/hai-mn/hai-mn.github.io, unless otherwise noted. The figures that have been reused from other sources don't fall under this license and can be recognized by a note in their caption: "Figure from ...".

Citation

For attribution, please cite this work as

Nguyen (2021, June 8). HaiBiostat: Metabolic Syndrome Prevalence across time using NHANES Data. Retrieved from https://hai-mn.github.io/posts/2021-06-09-mets-using-nhanesdata/

BibTeX citation

@misc{nguyen2021metabolic,
  author = {Nguyen, Hai},
  title = {HaiBiostat: Metabolic Syndrome Prevalence across time using NHANES Data},
  url = {https://hai-mn.github.io/posts/2021-06-09-mets-using-nhanesdata/},
  year = {2021}
}