Problem: Analyze the child obesity data with the model specified in the example and interpret the results.
Data: The data are from the Muscatine Coronary Risk Factor (MCRF) study, a longitudinal survey of school-age children in Muscatine, Iowa. The MCRF study had the goal of examining the development and persistence of risk factors for coronary disease in children. In the MCRF study, weight and height measurements of five cohorts of children, initially aged 5-7, 7-9, 9-11, 11-13, and 13-15 years, were obtained biennially from 1977 to 1981. Data were collected on 4856 boys and girls. On the basis of a comparison of their weight to age-gender specific norms, children were classified as obese or not obese. The goal of the analysis is to determine whether the risk of obesity increases with age and whether patterns of change in obesity are the same for boys and girls.
Column 1: ID
Column 2: Gender, 0 = Male, 1 = Female
Column 3: Baseline Age
Column 4: Current Age, Age denotes mid-point of age-cohort
Column 5: Occasions of measurements
Column 6: Obesity Status, 1 = Obese, 0 = Non-Obese, . = Missing
pacman::p_load(magrittr, dplyr, knitr, rmdformats, geepack, repolr, geesmv)
dta1 <- read.table("muscatine-data.txt", h=F, na.strings = '.')
names(dta1) <- c("ID","Gender","Age0","Age","Occasion","Obese")
head(dta1)
## ID Gender Age0 Age Occasion Obese
## 1 1 0 6 6 1 1
## 2 1 0 6 8 2 1
## 3 1 0 6 10 3 1
## 4 2 0 6 6 1 1
## 5 2 0 6 8 2 1
## 6 2 0 6 10 3 1
## 'data.frame': 14568 obs. of 6 variables:
## $ ID : int 1 1 1 2 2 2 3 3 3 4 ...
## $ Gender : int 0 0 0 0 0 0 0 0 0 0 ...
## $ Age0 : int 6 6 6 6 6 6 6 6 6 6 ...
## $ Age : int 6 8 10 6 8 10 6 8 10 6 ...
## $ Occasion: int 1 2 3 1 2 3 1 2 3 1 ...
## $ Obese : int 1 1 1 1 1 1 1 1 1 1 ...
dta1t <- dta1 %>%
mutate(Occasion = factor(Occasion, labels=c("1977", "1979", "1981"), ordered=T),
Gender = factor(ifelse(Gender == 0, "M", "F")),
Obese = factor(ifelse(Obese == 1, "Obese", "Non-Obese")))
ftable(dta1t, row.vars = c(2, 4), col.vars = c(5,6))
## Occasion 1977 1979 1981
## Obese Non-Obese Obese Non-Obese Obese Non-Obese Obese
## Gender Age
## F 6 141 23 0 0 0 0
## 8 294 58 265 55 0 0
## 10 270 92 276 87 268 90
## 12 291 91 279 99 278 92
## 14 300 89 226 64 256 73
## 16 0 0 250 87 226 56
## 18 0 0 0 0 140 37
## M 6 174 15 0 0 0 0
## 8 289 67 296 54 0 0
## 10 312 84 298 77 308 83
## 12 281 90 299 88 290 90
## 14 307 73 233 65 269 78
## 16 0 0 251 67 224 54
## 18 0 0 0 0 153 34
HARD to present in percentage for each occasion
par(mfrow=c(1,2))
barplot(prop.table(with(subset(dta1, Gender=="0"),
ftable(Obese, Age)), m=2),
xlab="Age \n (6, 8, 10, 12, 14, 16, 18 y/o)",
ylab="Proportion",
ylim=c(0,1),
main="Male",
beside=T)
legend('topleft', c("Non-Obese","Obese"),
col=c("black","gray"),
pch=15, bty='n', cex=.5)
barplot(prop.table(with(subset(dta1, Gender=="1"),
ftable(Obese, Age)), m=2),
xlab="Age \n (6, 8, 10, 12, 14, 16, 18 y/o)",
ylab="Proportion",
ylim=c(0,1),
main="Female",
beside=T)
The pattern is similar in male and female. However, male with higher non-obese proportion.
Model: Letting Yij = 1 if the ith child is classified as obese at the jth occasion, and Yij = 0 otherwise.
Assume that the marginal probability of obesity at each occasion follows the logistic model with age, gender, and gender × age effects, where age is defined by age at the jth occasion - 12 years.
It can be further assumed that the log odds of obesity changes curvilinearly with age (i.e., quadratic age trend), but the trend over time is allowed to be different for girls and boys.
All
dta1_m <- dta1 %>%
mutate(Age = Age - 12,
Age2 = Age^2)
summary(m0 <- geeglm(Obese ~ Gender + Age + Age2 + Gender:Age + Gender:Age2, data=dta1_m, id = ID, family = binomial, corstr = "ar1"))
##
## Call:
## geeglm(formula = Obese ~ Gender + Age + Age2 + Gender:Age + Gender:Age2,
## family = binomial, data = dta1_m, id = ID, corstr = "ar1")
##
## Coefficients:
## Estimate Std.err Wald Pr(>|W|)
## (Intercept) -1.205697 0.050734 564.783 < 2e-16 ***
## Gender 0.105306 0.071357 2.178 0.14001
## Age 0.042264 0.013433 9.899 0.00165 **
## Age2 -0.018092 0.003416 28.044 1.19e-07 ***
## Gender:Age 0.002900 0.018563 0.024 0.87585
## Gender:Age2 0.003495 0.004715 0.549 0.45857
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation structure = ar1
## Estimated Scale Parameters:
##
## Estimate Std.err
## (Intercept) 0.9943 0.02807
## Link = identity
##
## Estimated Correlation Parameters:
## Estimate Std.err
## alpha 0.6106 0.02036
## Number of clusters: 4856 Maximum cluster size: 3
omit the gender × age and gender × age2
summary(m1 <- geeglm(Obese ~ Gender + Age + Age2, data=dta1_m, id = ID, family = binomial, corstr = "ar1"))
##
## Call:
## geeglm(formula = Obese ~ Gender + Age + Age2, family = binomial,
## data = dta1_m, id = ID, corstr = "ar1")
##
## Coefficients:
## Estimate Std.err Wald Pr(>|W|)
## (Intercept) -1.21897 0.04788 648.17 < 2e-16 ***
## Gender 0.13145 0.06288 4.37 0.037 *
## Age 0.04387 0.00926 22.43 2.2e-06 ***
## Age2 -0.01630 0.00235 48.05 4.2e-12 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation structure = ar1
## Estimated Scale Parameters:
##
## Estimate Std.err
## (Intercept) 0.994 0.028
## Link = identity
##
## Estimated Correlation Parameters:
## Estimate Std.err
## alpha 0.61 0.0203
## Number of clusters: 4856 Maximum cluster size: 3
Obese | Obese | |||||
---|---|---|---|---|---|---|
Predictors | Odds Ratios | CI | p | Odds Ratios | CI | p |
(Intercept) | 0.30 | 0.27 – 0.33 | <0.001 | 0.30 | 0.27 – 0.32 | <0.001 |
Gender | 1.11 | 0.97 – 1.28 | 0.140 | 1.14 | 1.01 – 1.29 | 0.037 |
Age | 1.04 | 1.02 – 1.07 | 0.002 | 1.04 | 1.03 – 1.06 | <0.001 |
Age2 | 0.98 | 0.98 – 0.99 | <0.001 | 0.98 | 0.98 – 0.99 | <0.001 |
Gender * Age | 1.00 | 0.97 – 1.04 | 0.876 | |||
Gender * Age2 | 1.00 | 0.99 – 1.01 | 0.459 |
pacman::p_load(tidyr, ggplot2)
dta1_p <- dta1 %>%
mutate(Gender = factor(ifelse(Gender == 0, "Boys", "Girls")))
dta1p <- data.frame(dta1_p, id=row.names(dta1_p))
yhat_m0 <- data.frame(id=row.names(fitted(m0)),
phat=fitted(m0))
dta1_m0 <- inner_join(dta1p, yhat_m0, by="id")
ggplot(dta1_m0, aes(Age, group = Gender)) +
geom_line(data = dta1_m0, aes(y = phat, color = Gender)) +
labs(x="Age (years)",
y="Estimated Probability of Obesity") +
scale_y_continuous(limits = c(0.1, 0.27)) +
scale_x_continuous(limits = c(6, 18),
breaks = c(6, 8, 10, 12, 14, 16, 18),
minor_breaks = NULL) +
theme_minimal()+
theme(legend.position=c(.9, .2))
The estimated probability of obesity for boys at ages 6, 10, 14, and 18 is 0.12, 0.20, 0.23, and 0.18, respectively; for girls, the estimated probability of obesity at ages 6, 10, 14, and 18 is 0.13, 0.22, 0.26, and 0.20.
Source: Woolson, R.F., & Clarke, W.R. (1984). Analysis of categorical incompletel longitudinal data. Journal of the Royal Statistical Society, Series A, 147, 87-99.
Reported in Fitzmaurice, G.M., Laired, N.M., & Ware, J.H. (2004). Applied Longitudinal Data Analysis. pp. 306-312.
Reported in Fitzmaurice, G.M., Laired, N.M., & Ware, J.H. (2011). Applied Longitudinal Data Analysis. pp. 364-374.
Problem: Replicate the results of multiple sources analysis reported in the Cantabrian example.
Data: The data are from a survey of primary care patients in Cantabrian, Spain, to estimate the prevalence of psychiatric illness. The prevalence for each of the sexes is estimated from data provided by the general practitioner and from completing the general health questionnaire.
Column 1: Subject ID
Column 2: Illness = 1, Otherwise = 0
Column 3: Gender ID
Column 4: General practitioner = GP, General health questionnaire = GPQ
dta2 <- read.table("cantabrian.txt", h=T)
names(dta2) <- c("ID","Health","Gender","Type")
head(dta2)
## ID Health Gender Type
## 1 1 0 M GP
## 2 1 0 M GHQ
## 3 2 0 M GP
## 4 2 0 M GHQ
## 5 3 0 M GP
## 6 3 0 M GHQ
## 'data.frame': 1646 obs. of 4 variables:
## $ ID : int 1 1 2 2 3 3 4 4 5 5 ...
## $ Health: int 0 0 0 0 0 0 0 0 0 0 ...
## $ Gender: chr "M" "M" "M" "M" ...
## $ Type : chr "GP" "GHQ" "GP" "GHQ" ...
## Type GP GHQ
## Gender Health
## M Illness 287 244
## Otherwise 37 80
## F Illness 420 306
## Otherwise 79 193
## Type GP GHQ
## Gender Health
## M Illness 0.3487 0.2965
## Otherwise 0.0450 0.0972
## F Illness 0.5103 0.3718
## Otherwise 0.0960 0.2345
The prevalence in women is higher than in men.
par(mfrow=c(1,2))
barplot(prop.table(with(subset(dta2t, Type=="GP"),
ftable(Gender, Health)), m=2),
xlab="Health \n (Illness, Otherwise)",
ylab="Proportion",
ylim=c(0,1),
main="GP",
beside=T)
legend('topleft', c("Male","Female"),
col=c("black","gray"),
pch=15, bty='n', cex=.5)
barplot(prop.table(with(subset(dta2t, Type=="GHQ"),
ftable(Gender, Health)), m=2),
xlab="Health \n (Illness, Otherwise)",
ylab="Proportion",
ylim=c(0,1),
main="GHQ",
beside=T)
Multivariate binary responses
summary(m0 <- geeglm(Health ~ Gender + Type + Gender*Type, data=dta2, id = ID, corstr="exchangeable", family = binomial))
##
## Call:
## geeglm(formula = Health ~ Gender + Type + Gender * Type, family = binomial,
## data = dta2, id = ID, corstr = "exchangeable")
##
## Coefficients:
## Estimate Std.err Wald Pr(>|W|)
## (Intercept) -0.4609 0.0919 25.14 5.3e-07 ***
## GenderM -0.6542 0.1583 17.09 3.6e-05 ***
## TypeGP -1.2099 0.1265 91.46 < 2e-16 ***
## GenderM:TypeGP 0.2765 0.2283 1.47 0.23
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation structure = exchangeable
## Estimated Scale Parameters:
##
## Estimate Std.err
## (Intercept) 1 0.0786
## Link = identity
##
## Estimated Correlation Parameters:
## Estimate Std.err
## alpha 0.298 0.0474
## Number of clusters: 823 Maximum cluster size: 2
Health | |||
---|---|---|---|
Predictors | Odds Ratios | CI | p |
(Intercept) | 0.63 | 0.53 – 0.76 | <0.001 |
Gender [M] | 0.52 | 0.38 – 0.71 | <0.001 |
Type [GP] | 0.30 | 0.23 – 0.38 | <0.001 |
Gender [M] * Type [GP] | 1.32 | 0.84 – 2.06 | 0.226 |
The prevalence in women is higher than in men, irrespective of the case finder being used.
The GP detects fewer cases than the GHQ, irrespective of the subject’s gender.
Source: Everitt, B.S., & Dunn, G. (2001). Applied Multivariate Data Analysis, 2nd Ed. p. 237