The subjects came to baseline visits and were randomized between October 1994 and October 1995. The follow-up visits were exactly 12 and 36 months later to avoid the effects of seasonal changes. For ...