Strategies to meet the need for long-term data.

Chronic diseases afflict patients for many years, often to the end of life, and there is increasing need for estimating lifelong risk and for evaluating the effects of treatment in the long term. Yet recommendations for lifelong treatment are most frequently based on findings from randomized clinical trials lasting only a few years. There is therefore a clear need for much longer term data, and here we present the advantages and disadvantages of many strategies, including the use of long-term posttrial follow-up, of long-term prospective cohort studies, registry databases, and of administrative databases. We also emphasize the need for long-term cost-effectiveness studies. One of the most promising strategies comes from linkage of data gathered through the ever-expanding pool of administrative databases worldwide with data from other sources, including randomized trials and the many forms of observational study.