Data selection and preparation

Data input:

The dataset that is used for this analysis contains 16 patients with measurements for weight and triglyceride levels that are taken before and after a diet intervention. The following variables are selected for the analysis:

• Pre-diet triglyceride levels (tg0) and Post-diet triglyceride levels (tg4)
• Pre-diet weight (wgt0) and Post-diet weight (wgt4)

Criteria subdialog box:

• Scope: All available data was used with pairwise deletion. This ensures that any missing data is handled by excluding cases where values are missing only for the specific comparison
• Agreement Limits: The default settings were used, specifying the limits of agreement at 95% percentile.
• Confidence interval of agreement limits and Confidence interval of mean differences: Both the agreement limits and mean differences were calculated with 95% confidence intervals, providing an estimation of the variability in these metrics.

Print subdialog box

The Bland-Altman print subdialog box allows to customize the output of your analysis.

Mean Differences and Confidence Intervals

Mean Difference

The mean difference provides an estimate of the bias between the two measurement methods. In a Bland-Altman plot, this is represented by the central horizontal line. Selecting this option includes the mean difference in the printed output.

Confidence Intervals for Mean Difference

These intervals indicate the range within which the true mean difference is likely to fall, typically at a 95% confidence level. Including these intervals in the output helps you understand the precision of the mean difference estimate. A mean difference close to zero with narrow confidence intervals suggests good agreement between the two methods. A wide confidence interval implies less precision in the estimate of the mean difference.

Agreement Limits and Confidence Intervals

Limits of Agreement

The limits of agreement define the range within which most differences between the two methods lie, typically covering 95% of the data points. In the plot, these are shown as two horizontal lines above and below the mean difference.

Confidence Intervals for Limits of Agreement

These intervals provide a range for each of the limits of agreement, reflecting the uncertainty around these limits. Including these in the output is crucial for assessing the reliability of the agreement limits.

Narrow limits of agreement suggest that the two methods have consistent differences, indicating good agreement. Wide limits may indicate poor agreement, with substantial variation in the differences between the two methods.

Print Details of Analysis

Selecting this option prints a comprehensive summary of the Bland-Altman analysis, including all relevant statistics such as mean difference, limits of agreement, and their respective confidence intervals. This includes an assessment of the distribution of differences, helping you to identify any outliers or patterns that might affect the interpretation of agreement.

With the plot sub dialog box you can configure the visual elements of your Bland-Altman plots.

Choosing the "Ratio Plot" option generates plots that display the ratio of the differences to the mean of the measurements, expressed as a percentage. This visualization is useful when the variability between measurements is proportional to their magnitude. By showing relative differences, ratio plots can highlight proportional bias that might not be as evident in traditional difference plots.

Y-axis represents the ratio of differences to the mean, either as a raw ratio or in percentage form. The y-axis can be labeled as (Variable 1 - Variable 2) / Mean for ratios or (Variable 1 - Variable 2) / Mean % for percentages, depending on the user’s preference.

X-axis represents the mean of the two measurements, allowing for the examination of the relationship between the size of the measurements and their relative differences. Ratio plots help visualize agreement between the two methods in relative terms. If the points are consistently around a specific percentage (for example, close to 0%), it indicates that the methods agree relatively across the measurement range. Deviations from this line can indicate systematic differences that vary with the magnitude of the measurements.

Select Agreement Limits to add lines to the plot that denote the range within which 95% of the ratios are expected to lie. In ratio plots, these limits are expressed as percentages above and below the mean difference. This feature helps you understand the expected range of variation in the relative differences

Upper and lower limits represent the bounds within which most ratios are expected to fall, indicating the range of acceptable agreement between the two methods. Lines are drawn at these limits on the plot, providing a clear visual reference for assessing how well the measurements agree in relative terms.

Select Confidence Intervals of Mean Difference to include confidence intervals for the mean difference (or ratio) in the plot. This addition provides a visual representation of the uncertainty around the estimated mean difference or ratio. A solid line representing the mean of the ratio differences and dashed lines around the mean difference line, indicate the range in which the true mean difference is expected to lie with a specified level of confidence. Points that are scattered around the mean difference line provide insight into the agreement between the two methods in relative terms. If the points lie within the confidence intervals and the agreement limits, it suggests that the methods agree within an acceptable range. The confidence intervals also help assess the precision and reliability of the estimated mean difference, indicating the degree of certainty in the observed agreement.