Bosenberg's (1995) Paper and SED Data

Notes

Bosenberg, A. T. (1995) "Skin-epidural distance in children." Anaesthesia. 50: 895-897.

• SED Data (digitized from paper figure): Bosenberg1995.csv
• Scanned image from the Bosenberg (1995) paper: Bosenberg1995.png
• Digitized SED figure from the Bosenberg paper (Fig. 2): Bosenberg1995.svg
• Fitted linear regression line: SED = 3.93 + 0.80 * WT, where SED is in mm units, and WT is in kg units (R² = 0.74).
• From the paper:

Statistical methods: Regression analysis using the least squares procedure was used to determine the association between the skin-epidural distance and weight. For the difference between the two equations, i.e. the regression equation and the hypothesised 1 mm = 1 kg body weight, a Z-test was performed.

In children over 10 years (n = 19) there was also poor correlation. The data from these patients were not used for further analysis.

Plotting the best fit regression line and its 95% confidence limits shows that the y = x line, or skin-epidural distance (mm) is equal to weight (kg) falls within these limits for weight <40 kg (or 10 years of age) in this series (Fig. 2). Therefore the difference between any two points on these two lines is not statistically significant.

In addition a Z-test performed on any two points obtained from the two equations for a specific weight, i.e. skin-epidural distance (mm) = 0.80 weight (kg) 3.93, and skin-epidural distance (mm) = weight (kg) was found to be not statistically significant.

...

In summary, we have demonstrated that 1 mm/kg is a useful guideline for judging the skin-epidural distance in children between 6 months and 10 years of age.

Discussion Points

• Was the confidence band really a prediction band?
• Was it simultaneous or pointwise?
• Was Bosenberg's reasoning sound?
• How can we verify Bosenberg's conclusions?

Assignment

Modify the Bosenberg Sweave file from the previous assignment as follows, and regenerate the PDF report. Ensure that the document remains reproducible.

1. Add Bosenberg's simplified prediction rule (i.e., 1 mm/kg) to the scatterplot.
2. Modify the figure caption to reflect the addition above.
3. Use a statistical technique to assess whether Bosenberg's simplified rule was reasonable.
4. Add a short discussion of your findings.

Solution

Here is one solution to the assignment above: Bosenberg.Rnw Bosenberg.pdf