At the end of the trial 3721 women with an intact uterus or subtotal hysterectomy had been randomised, 1862 to combined HRT and 1859 to placebo.13
The figure shows the flow of women after randomisation for this report. We excluded women who had been in the trial less than 40 weeks at trial closure in October 2002 or who had died within one year of randomisation (total 671 combined HRT; 666 placebo). Of the remainder, 148 and 106 women randomised to combined HRT and placebo, respectively, did not attend their one year interview and cannot contribute to the primary analyses (figure). Mean age at randomisation in the 2130 women in the primary analyses was 63.8 years (SD 4.4). The median time between randomisation and one year interview was 368 days (90% interviews were 331-456 days after randomisation), and 21% of one year interviews (21% combined HRT; 20% placebo) were completed after trial closure.
Flow of participants after randomisation
At the first annual medical 290 (28%) women randomised to combined HRT were no longer taking trial treatment (214 had stopped permanently and 76 were on a temporary interruption of treatment). Of the 214 who stopped combined HRT permanently, 91 did so in the first 14 weeks and 177 did so in the first 27 weeks. Overall, trial treatment was supplied for 79% of follow-up time between randomisation and first annual medical. Correspondingly, at the first annual medical, 141 (13%) in the placebo group were not taking trial treatment (89 had stopped permanently; 52 were on a temporary interruption); trial treatment was supplied for 92% of follow-up time. The most common reasons for permanent discontinuation of combined HRT were vaginal bleeding (68/214, 32%) and breast tenderness (28/214, 13%). Of the patients who were not taking trial treatment at the first annual medical, 174 (60%) in the combined HRT group and 10 (7%) in the placebo group had previously reported bleeding to some degree, including spotting. Of those taking trial treatment at one year, 265 (37%) in the combined HRT group had reported bleeding at some time during the first year, including 65 (9%) since their last visit. Corresponding proportions in the placebo group were 36 (4%), including 9 (1%) since their last visit.
We reviewed the data on the women who did not attend their one year interview and therefore could not be included in the primary analyses. Most were known to have discontinued trial treatment: 132/148 (89%) women randomised to combined HRT stopped treatment permanently before one year (n=118) or were on a temporary interruption of treatment at one year (n=14) and did not restart treatment before study closure; in the placebo group 84/106 (79%) had discontinued trial treatment (68 permanently and 16 on a temporary interruption). As in the women who attended the one year interview, the most common reasons for permanent discontinuation of combined HRT were vaginal bleeding (42/118, 36%) and breast tenderness (13/118, 11%.).
Women’s health questionnaire
According to assessment with the women’s health questionnaire, participants in the combined HRT group experienced significant improvements in vasomotor symptoms compared with the placebo group at one year (difference between treatment groups, adjusted for baseline score, 0.09, 95% confidence interval 0.07 to 0.12), P<0.001) (table 1). Treatment differences were more marked in those with more severe baseline symptoms (P<0.001). Participants also reported small but significant improvements related to sexual functioning (difference between treatment groups, adjusted for baseline score, 0.05, 0.02 to 0.08, P<0.001) and sleep problems (difference between treatment groups, adjusted for baseline score, 0.05, 0.02 to 0.07, P<0.001). Although sleep problems at one year were marginally higher in those with baseline vasomotor symptoms, there was no evidence for a difference in treatment benefit by baseline vasomotor symptoms (P=0.78). Sexual function at one year was not associated with baseline vasomotor symptoms (P=0.78).
Table 1 Mean (SE) scores on health related quality of life as measured with women’s health questionnaire by treatment group
Table 2 shows the distribution of individual menopausal symptoms by treatment group. At one year, with adjustment for differences at baseline, participants randomised to combined HRT experienced fewer vasomotor symptoms, including hot flushes (9% v 25%, P<0.001) and night sweats (14% v 23%, P<0.001) than those randomised to placebo. They were also less likely to report symptoms of aching joints and muscles (57% v 63%, P=0.001), insomnia (35% v 41%, P<0.001), and vaginal dryness (14% v 19%, P<0.001). Bloating was marginally less prevalent in the combined HRT group than the placebo group (21% v 24%, P=0.005), but significance was not achieved when we allowed for multiple testing. The combined HRT group reported higher rates of breast tenderness (16% v 7%, P<0.001) and vaginal discharge (14% v 5%, P<0.001) than the placebo group.
Table 2 Prevalence of symptoms related to menopause. Figures are numbers (percentages) of women
In a secondary analysis of symptoms at one year we carried forward the last observation after entry for all women without a one year interview who had been in the trial at least 40 weeks at closure (allowing us to include 99% of women in both treatment groups). Overall results (not shown) were similar, although the proportion of women reporting breast tenderness in the combined HRT group at their last visit who had not attended a one year interview was somewhat higher (48%, 64/133) than in the group interviewed at one year (16%, 164/1043).
Vasomotor symptom interactions
We explored interactions between the treatment effect at one year and baseline reporting of hot flushes or night sweats. The benefit of combined HRT was greater in those who had symptoms at baseline: at one year in the combined HRT group 22% of those who had reported hot flushes at baseline were still having hot flushes compared with 65% in the placebo group, while in women who had not reported hot flushes at baseline corresponding proportions were 4% and 9%, respectively (interaction P=0.002). Similarly, only 33% of women randomised to combined HRT who reported night sweats at baseline were still having them at one year compared with 62% of women randomised to placebo; corresponding proportions of those who had not reported night sweats at baseline were 7% and 10% (interaction P=0.002). We investigated whether the impact of combined HRT on other symptoms varied according to baseline vasomotor symptoms but found no interaction for aching joints and muscles, insomnia, vaginal dryness, breast tenderness, or vaginal discharge.
Assessment of depression with the CES-D at one year showed no significant difference between the combined HRT (median 3, interquartile range 0-7) and placebo groups (median 3, 1-8; P=0.51 for difference, adjusted for baseline score) and no difference in the proportion of individuals who experienced high depressive scores (CES-D >16 units) between the two treatment groups (8% v 9%; P=0.51, adjusted for baseline).
Quality of life
There were no differences in self assessed health measured by the EQ visual analogue scale at one year follow-up (table 3). Participants randomised to combined HRT had a marginally higher health classification index score compared with those randomised to placebo (difference between treatment groups adjusted for baseline score 0.016 units, 95% confidence interval 0.003 to 0.028, P=0.02) but the difference was not significant when we applied the Bonferroni correction.
Table 3 EuroQoL scores by treatment group. Figures are means (SE)
Generic visual analogue scale
At four weeks women in the combined HRT group had a slightly lower overall quality of life than those in the placebo group (difference, adjusted for baseline values, −1.6 units, −2.7 to −0.4, P=0.006, table 4). This difference was reduced at 14 weeks (P=0.03), and there were no differences between the combined HRT and the placebo groups after 14 weeks. We lacked power to investigate the extent to which the early reduction in visual analogue score associated with combined HRT declined because women who experienced a negative impact of combined HRT on quality of life stopped trial treatment early. The difference between treatment groups at four weeks was somewhat greater in women who stopped trial treatment before their first annual medical (difference −2.6 units, −5.4 to 0.2) than in women who were on trial treatment at one year (difference −1.0 units, −2.2 to 0.2, P=0.3 for difference).
Table 4 Differences in generic visual analogue scale for overall quality of life between treatment groups
We repeated analyses using all available data at each interview (on participants randomised at least 40 weeks before closure) and found similar results (treatment difference at four weeks −1.35 units, P=0.012; 1137 patients in the combined HRT group and 1141 patients in the placebo group).