Biostatistics · Intermediate

Case Study: Effects of Tobacco Use on Bone-Mineral Density (BMD) in Middle-Aged Women

Written by SignalsLab

Early edition: these lessons are being reviewed.

Content

Summary

This lesson examines a 1994 study by Hopper and Seeman that used a matched-pair design of 41 female twin pairs to isolate the effect of tobacco use on bone-mineral density (BMD). By comparing heavier-smoking twins to their lighter-smoking counterparts, the study demonstrated a statistically significant inverse relationship between smoking intensity and BMD, particularly at the lumbar spine and femoral neck.

Intuition

The core challenge in observational studies is distinguishing the effect of a specific exposure, like smoking, from other confounding variables, such as genetics. This study elegantly solves this by using monozygotic twins, who share nearly identical genetic backgrounds. By pairing a heavier-smoking twin with a lighter-smoking twin, researchers effectively cancel out genetic influences on bone density. The analysis then focuses solely on the 'within-pair' differences. If the heavier smoker consistently has lower BMD than their twin, and this difference correlates with the difference in pack-years, we can attribute the bone loss to tobacco use rather than heredity. This design transforms a complex population-level question into a controlled comparison of differences.

Key equations

t=x‾d−μ0SE
(1)

The t-statistic formula for a one-sample t-test on paired differences, where x‾d is the mean difference, μ0 is the hypothesized mean difference (0), and SE is the standard error.

Derivation

  1. 1. Define the variable of interest: The difference in BMD between the heavier-smoking twin and the lighter-smoking twin for each pair.
  2. 2. Standardize the exposure: Quantify tobacco use in pack-years (1 pack/day for 1 year).
  3. 3. Calculate within-pair differences: Subtract the lighter-smoker's BMD from the heavier-smoker's BMD to eliminate between-subject variability.
  4. 4. Formulate hypotheses: Null hypothesis (H0) states the mean difference μd=0; Alternative hypothesis (H1) states μd≠0.
  5. 5. Compute the test statistic: Use the formula t=x‾d/ SE, where SE is the standard error of the mean difference.
  6. 6. Determine significance: Compare the calculated t-value against the t-distribution with n-1 degrees of freedom (40 df) to find the p-value.

Worked example

Problem

Calculate the t-statistic and interpret the significance for the lumbar spine BMD data, where the mean percentage difference is -5.0% and the standard error is 2.0%.

  1. Identify the mean difference (x‾d) as -5.0%.
  2. Identify the standard error (SE) as 2.0%.
  3. Assume the null hypothesis mean difference (μ0) is 0.
  4. Apply the t-statistic formula: t = (-5.0 - 0) / 2.0.
  5. Calculate the result: t = -2.5.
  6. Compare this t-value to the critical values for 40 degrees of freedom. The value -2.5 falls between the critical thresholds for alpha=0.02 (approx 2.423) and alpha=0.01 (approx 2.704).
  7. The precise p-value is 0.017, which is less than 0.02, indicating statistical significance.

Answer The t-statistic is -2.5. With a p-value of 0.017, the result is statistically significant, confirming that heavier smokers have significantly lower BMD at the lumbar spine than their lighter-smoking twins.

Common mistakes

  • Confusing the direction of subtraction: The study defines the difference as heavier-smoker BMD minus lighter-smoker BMD, resulting in negative values for bone loss.
  • Ignoring the matched-pair design: Treating the 82 individuals as independent samples rather than 41 paired differences violates the study's statistical assumptions.
  • Misinterpreting pack-years: Failing to recognize that pack-years standardize both intensity and duration of smoking.
  • Overgeneralizing results: The study specifically involved middle-aged Australian women; results may not apply to other demographics without further study.

Animation

Animation coming soon

The animated version of this lesson is being prepared for the web. In the Signals AI app, lessons are drawn step by step.

Code

This is a text lesson: it has no animation code yet.

ccBYNC© 2026 SignalsLab. Lesson text shared as is under CC BY-NC 4.0.