Digitalis Investigation Group (DIG) "Teaching" Data Set

SAS code to access the data,   Proc Print output,   Proc Means output.

More complete description in pdf or in MS Word.

Brief Description

The Digitalis Investigation Group (DIG) Trial was a randomized, double-blind, multicenter trial with more than 300 centers in the United States and Canada participating. The purpose of the trial was to examine the safety and efficacy of Digoxin in treating patients with congestive heart failure in sinus rhythm. Digitalis was introduced clinically more than 200 years ago and has since become a commonly prescribed medication for the treatment of heart failure; however, there was considerable uncertainty surrounding its safety and efficacy. Small trials indicated that Digoxin alleviated some of the symptoms of heart failure, prolonged exercise tolerance, and generally improved the quality of patients' lives. Unfortunately, these trials were generally small and although they did focus on the effect of treatment on patients' relief from heart failure symptoms and quality of life, they failed to address the effect of treatment on cardiovascular outcomes. Questions about the safety of Digoxin were also a concern. Digoxin toxicity is uncommon in small trials with careful surveillance, however, the long-term effects of therapeutic levels of Digoxin were less clear.

The DIG dataset consists of baseline and outcome data from the main DIG trial. In the main trial, heart failure patients meeting the eligibility criterion and whose ejection fraction was 45% or less were randomized to receive either a placebo or digoxin. Outcomes assessed in the trial included: cardiovascular mortality, hospitalization or death from worsening heart failure, hospitalization due to other cardiovascular causes and hospitalization due to non-cardiovascular causes. The dataset was prepared to enable students to reproduce the analysis leading to the results of the paper published in the New England Journal of Medicine in 1997:

"The effect of digoxin on mortality and morbidity in patients with heart failure," The Digitalis Investigation Group, N Engl J Med. 1997 Feb 20;336(8):525-33.

Some data not discussed in the NEJM article are included in the teaching data set (body mass index, serum creatinine, serum potassium, systolic and diastolic blood pressure, etc.). In order to create an anonymous dataset that protects patient confidentiality, most variables have been permuted over the set of patients within treatment group. Therefore, this dataset can reproduce the results of the NEJM paper; however, it would be inappropriate to use this dataset for other research or publication purposes. Multidimensional relationships, not included in the NEJM, may not have been preserved during the permutation process.