PH784: Problem Set #1 (due Sept 15, 2025)

Author

Koichiro Shiba

Published

September 2, 2025

Intro

We will analyze the 2009-2012 NHANES data via the R package NHANES. This document provides a comprehensive info about the package. This package provides two datasets: NHANES and NHANESraw. In the documentation, the following is stated:

NHANES and NHANESraw each include 75 variables available for the 2009-2010 and 2011-2012 sample years. NHANESraw has 20,293 observations of these variables plus four additional variables that describe that sample weighting scheme employed. NHANES contains 10,000 rows of data resampled from NHANESraw to undo these oversampling effects. NHANES can be treated, for educational purposes, as if it were a simple random sample from the American population.

We will use the dataset NHANES (n=10,000) and ignore sampling weights.

  • Outcome
    • Continuous: BMI, BPSysAve
    • Binary: Diabetes, SmokeNow
  • Social group
    • Race: Race1
    • Educational attainment: Education
    • Income: HHIncome

1. Pair-wise comparison

1.1 Use the pair-wise comparison approach to quantify the education-based disparity in the prevalence of current smoker in the NHANES dataset. Report your estimate and show your R code.

1.2 What is your reference group?

1.3 What is the scale of the inequality you estimated? Also, briefly explain how you would estimate inequalities in the other scale.

2. Mean difference measures

2.1 The table below shows mean systolic blood pressure levels across the education groups among individuals with no missing data in the two variables.

Education n mean.SBP
8th Grade 423 125.9220
9 - 11th Grade 866 122.3349
High School 1460 122.4103
Some College 2175 121.0143
College Grad 2036 118.3777

The overall systolic blood pressure was (see code output). Calculate the mean difference from a reference group for the absolute education-based inequality in systolic blood pressure. Report your estimate and show your R code.

2.2 Which group did you chose as a reference group?

2.3 The R package healtheqaul allows you to compute health disparities easily. Here is the tutorial document for the package. Using the default settings for the function mdmu(), would you expect to obtain the identical estimate for the absolute inequality? Why? Why not? You do not need to actually calculate the disparity using the package. Simply explain your reasoning. Read this page and see how the function handles a reference group and weighting.

3. Regression-based Measures

Read this NEJM article by Mackenbach et al. (2008). Table 2 shows estimated SIIs for the educaiton-based absolute inequality across European countries.

3.1 Interpret the estimated SII for all-cause mortality among men in Sweden (625) and in Hungary (2580). Based on these results, which country do you think is experiencing greater education-based inequality in all-cause mortality?

3.2 Explain the key assumption underlying their use of SII as a measure of socioecnomic inequality in health and discuss whether the authors provided any justification to support the validity of the assumption. If not, briefly comment on how you would check whether the assumption is valid.

Quarto Tip

When providing your answers, use callout blocks for better readability.

Put your answer inside a callout block. I suggest using the appearance="minimal" for the clean appearance.