Home Economics

Example Datasets

The Age-Related Macular Degeneration (ARMD) Trial

The objective of this trial was to examine the efficacy of interferon-a to treat age-related macular degeneration (ARMD). ARMD is a medical condition in which patients progressively lose vision (Pharmacological Therapy for Macular Degeneration Study Group, 1997). In the ARMD trial, visual acuity was examined using standardized vision charts that display lines with five letters of decreasing sizes (see Figure 2.1). The patients had to read these letters from top (largest letters) to bottom (smallest letters). Visual acuity was quantified as the total number of letters that were correctly read by a patient.

A total of 181 patients from N = 36 centers participated in the ARMD trial. There were two treatment conditions: interferon-a and placebo (coded as 1 = interferon-a and — 1 = placebo). The true endpoint (T) is the change in visual acuity 52 weeks after the start of the treatment. The candidate surrogate endpoint (S) is the change in visual acuity 24 weeks after starting the treatment. A total of 84 and 97 patients were enrolled in the placebo and interferon-a treatment conditions, respectively.

Figure 2.2 shows histograms of the changes in visual acuity after 24 and 52 weeks. As can be seen, both endpoints are roughly normally distributed. Figure 2.3 presents a scatter plot of T against S. There was a moderately strong correlation between S and T in both the placebo and the experimental treatment groups, i.e., r(S, T) = 0.7693 and 0.7118, respectively. The overall correlation combining the information of both treatment groups was r(S, T) = 0.7464.

The aim of the ARMD trial was to show that interferon-a is superior to the placebo treatment (using visual acuity as the primary endpoint). Recall that

FIGURE 2.1

Age-Related Macular Degeneration Trial. Vision chart.

for a (two-sided) superiority trial, the null and the alternative hypotheses are H0 : Д = 0 and Ha : Д = 0 (with Д= the true difference in the means of the outcomes for both treatment groups). When the result of the hypothesis test is significant, it is concluded that the effect of the experimental treatment on the outcome differs from the effect of the control treatment on the outcome. When the observed result is in favor of the experimental treatment, it is concluded that the experimental treatment is performing significantly better as compared to the control treatment. Note that in case of a non-significant result, it cannot be concluded that the experimental treatment is equally good as the control treatment. The reason for this is that for each small true difference Д, one can always establish a sample size for which H0 will be rejected with a high probability (Lesaffre, 2008).

In the ARMD dataset, the means of the changes in visual acuity after 24 weeks (S) were -6.0309 in the placebo group and -7.8095 in the interferon- a group (Дs = -1.7786, t = -0.9352, p = 0.3509; two-sided t-test). The means of the changes in visual acuity after 52 weeks (T) were -11.7423 in the placebo group and -14.6548 in the interferon-a group (Дт = -2.9125, t = -1.2372, p = 0.2176; two-sided t-test). There was thus no evidence that the true differences in the means of S and T in both treatment groups differed from zero.

In the remainder of this book, this dataset will be referred to as the ARMD

FIGURE 2.2

Age-Related Macular Degeneration Trial. Histograms of the change in visual acuity 24 weeks after starting the treatment (the candidate surrogate; left panel) and 52 weeks after starting the treatment (the true endpoint; right panel).

dataset. The ARMD dataset is included in the R library Surrogate (where it can be accessed using the command data(ARMD)). Alternatively, it can be downloaded from http://ibiostat.be/online-resources (file ARMD.txt) for use in SAS or any other software package. Table 2.1 provides a summary of the variables that are included in the ARMD dataset. The dataset is organized in the “wide” format (i.e., there is one line of data per patient). By means of illustration, Table 2.2 shows the first 5 observations in the ARMD dataset.

The data were provided by the Pharmacological Therapy for Macular Degeneration Study Group (Pharmacological Therapy for Macular Degeneration Study Group, 1997).

 Related topics