Background and context

In a recently published paper, Toh and colleagues re-examined data from the Women's Health Initiative (WHI) estrogen plus progesterone (E+P) randomised trial1 asking whether the apparent increase in risk of coronary heart disease (CHD) events observed in the first 2 years of this trial was a time-dependent phenomenon and whether it might dissipate with prolonged treatment. Although the WHI E+P trial was stopped short of the planned 8.5-year duration, there were still 11,530 women of the 16,485 enrolled for whom 5 years of observations were available and 9,372 for whom 6 or more years of observations could be examined, making it feasible to address this question.

In a subsequent report containing a more complete analysis of adjudicated CHD outcomes from the WHI E+P trial2, Hazard Ratios (HR's) for CHD appeared to decrease from a high point in year 1, at 1.81 (95% CI 1.09–3.01), to a low point, at 0.70 (0.42–1.14), over a 6-year follow-up. During years 2 through 5, HR's were greater than 1.0, ranging from 1.34 to 1.45. However, only the year-1 HR was statistically significant. In their discussion, Manson and colleagues cautioned that the apparent trend for decreasing risk should be interpreted conservatively because there were relatively fewer events and subjects and lower rates of adherence to study …