Hormone Study Conflicts With Results From WHI

RENO, NEV. — Women taking hormone therapy had no increase in cardiovascular events and a lower overall death rate, compared with age-matched controls, in a retrospective study that used a primary care database and that was designed to mimic the patient population enrolled in the Women's Health Initiative.

The findings stand in sharp contrast to results of the WHI, although the investigators set out to match the cohort to the WHI study population in terms of inclusion criteria, study time frame, treatment, and outcome variables.

Dr. Kurt T. Barnhart and his associates at the University of Pennsylvania, Philadelphia, examined the records of women in the United Kingdom General Practice Research Database (GPRD), including 13,658 women aged 55–79 years who were taking combination estrogen/progestin hormone therapy (HT), 37,730 matched controls, and a separate group of younger subjects: 20,654 women on HT and 30,102 controls aged 50–55 years.

The results were presented in poster form at the annual meeting of the Society for Gynecologic Investigation.

“We found, in contrast to WHI, that there was no cardioadverse association with hormone replacement therapy. We didn't find it cardioprotective, either. In other words, if this had been the WHI, it wouldn't have been stopped,” explained Dr. Barnhart, who is with the department of obstetrics and gynecology at the university.

Women were selected for the retrospective study if their demographics matched those of the WHI cohort. They either took 0.625 mg daily of conjugated estrogen and 150 mcg of norgestrel on days 17–28 per cycle, or they served as controls.

In the GPRD, the adjusted hazard ratio for myocardial infarction was 0.95 (0.78–1.16) for older women and 0.91 (0.69–1.20) for younger women.

By contrast, in the WHI, the hazard ratio for nonfatal MI was 1.28 (0.96–1.70); for coronary heart disease deaths (including fatal MI), it was 1.10 (0.65–1.89) (N. Engl. J. Med. 2003;349:523–34).

Death from all causes was significantly lower in the older GPRD subjects taking HT than in control subjects, with a hazard ratio of 0.75 (0.65–0.86). It also was lower among the younger women taking hormones vs. younger controls (hazard ratio 0.76 [0.63–0.91]). In the WHI, overall mortality was not affected by HT.

In an interview at the meeting, Dr. Barnhart called the reduced mortality finding “mildly surprising.”

“Part of me wants to say that death is really the only thing that matters,” he said.

“You could have a stroke or have breast cancer or you could be protected from colorectal cancer, but really it matters what happened to you, and it looked like there was a lower death rate. I don't know why.”

Analyses are underway to determine whether missing data may help to account for the mortality findings that differed from the WHI results.

In some respects, the GPRD study closely paralleled WHI conclusions. For example, elevated rates of stroke were seen in both studies, with hazard ratios of 1.23 in the GPRD and 1.41 in WHI results that were reported after the trial was discontinued early (JAMA 2002;288:321–33). Similarly, breast cancer was elevated among hormone users in both trials, with hazard ratios of 1.67 and 1.26 in the GPRD and WHI, respectively.

Venous thromboembolic events were elevated in hormone users in both trials, with hazard ratios of 1.55 in the GPRD and 2.22 in the WHI, while the risk of colorectal cancer was diminished (with a hazard ratio of 0.56 in the GPRD and 0.63 in the WHI).

The hazard ratios for hip fractures were similar in both studies, 0.82 in the GPRD and 0.66 in the WHI.

Dr. Barnhart said the sheer size and scope of the GPRD, plus the meticulous inclusion criteria, served to overcome bias often associated with observational studies as opposed to randomized trials such as the WHI.

“If you have a relatively small observational study, then you've got the possibility of selecting women that might be healthier. But when you look at a large number of women, a cross section of the population—this is close to 10% of the population—you're much more likely to see what's actually happening in real life.”

RENO, NEV. — Women taking hormone therapy had no increase in cardiovascular events and a lower overall death rate, compared with age-matched controls, in a retrospective study that used a primary care database and that was designed to mimic the patient population enrolled in the Women's Health Initiative.

The findings stand in sharp contrast to results of the WHI, although the investigators set out to match the cohort to the WHI study population in terms of inclusion criteria, study time frame, treatment, and outcome variables.

Dr. Kurt T. Barnhart and his associates at the University of Pennsylvania, Philadelphia, examined the records of women in the United Kingdom General Practice Research Database (GPRD), including 13,658 women aged 55–79 years who were taking combination estrogen/progestin hormone therapy (HT), 37,730 matched controls, and a separate group of younger subjects: 20,654 women on HT and 30,102 controls aged 50–55 years.

The results were presented in poster form at the annual meeting of the Society for Gynecologic Investigation.

“We found, in contrast to WHI, that there was no cardioadverse association with hormone replacement therapy. We didn't find it cardioprotective, either. In other words, if this had been the WHI, it wouldn't have been stopped,” explained Dr. Barnhart, who is with the department of obstetrics and gynecology at the university.

Women were selected for the retrospective study if their demographics matched those of the WHI cohort. They either took 0.625 mg daily of conjugated estrogen and 150 mcg of norgestrel on days 17–28 per cycle, or they served as controls.

In the GPRD, the adjusted hazard ratio for myocardial infarction was 0.95 (0.78–1.16) for older women and 0.91 (0.69–1.20) for younger women.

By contrast, in the WHI, the hazard ratio for nonfatal MI was 1.28 (0.96–1.70); for coronary heart disease deaths (including fatal MI), it was 1.10 (0.65–1.89) (N. Engl. J. Med. 2003;349:523–34).

Death from all causes was significantly lower in the older GPRD subjects taking HT than in control subjects, with a hazard ratio of 0.75 (0.65–0.86). It also was lower among the younger women taking hormones vs. younger controls (hazard ratio 0.76 [0.63–0.91]). In the WHI, overall mortality was not affected by HT.

In an interview at the meeting, Dr. Barnhart called the reduced mortality finding “mildly surprising.”

“Part of me wants to say that death is really the only thing that matters,” he said.

“You could have a stroke or have breast cancer or you could be protected from colorectal cancer, but really it matters what happened to you, and it looked like there was a lower death rate. I don't know why.”

Analyses are underway to determine whether missing data may help to account for the mortality findings that differed from the WHI results.

In some respects, the GPRD study closely paralleled WHI conclusions. For example, elevated rates of stroke were seen in both studies, with hazard ratios of 1.23 in the GPRD and 1.41 in WHI results that were reported after the trial was discontinued early (JAMA 2002;288:321–33). Similarly, breast cancer was elevated among hormone users in both trials, with hazard ratios of 1.67 and 1.26 in the GPRD and WHI, respectively.

Venous thromboembolic events were elevated in hormone users in both trials, with hazard ratios of 1.55 in the GPRD and 2.22 in the WHI, while the risk of colorectal cancer was diminished (with a hazard ratio of 0.56 in the GPRD and 0.63 in the WHI).

The hazard ratios for hip fractures were similar in both studies, 0.82 in the GPRD and 0.66 in the WHI.

Dr. Barnhart said the sheer size and scope of the GPRD, plus the meticulous inclusion criteria, served to overcome bias often associated with observational studies as opposed to randomized trials such as the WHI.

“If you have a relatively small observational study, then you've got the possibility of selecting women that might be healthier. But when you look at a large number of women, a cross section of the population—this is close to 10% of the population—you're much more likely to see what's actually happening in real life.”

RENO, NEV. — Women taking hormone therapy had no increase in cardiovascular events and a lower overall death rate, compared with age-matched controls, in a retrospective study that used a primary care database and that was designed to mimic the patient population enrolled in the Women's Health Initiative.

The findings stand in sharp contrast to results of the WHI, although the investigators set out to match the cohort to the WHI study population in terms of inclusion criteria, study time frame, treatment, and outcome variables.

Dr. Kurt T. Barnhart and his associates at the University of Pennsylvania, Philadelphia, examined the records of women in the United Kingdom General Practice Research Database (GPRD), including 13,658 women aged 55–79 years who were taking combination estrogen/progestin hormone therapy (HT), 37,730 matched controls, and a separate group of younger subjects: 20,654 women on HT and 30,102 controls aged 50–55 years.

The results were presented in poster form at the annual meeting of the Society for Gynecologic Investigation.

“We found, in contrast to WHI, that there was no cardioadverse association with hormone replacement therapy. We didn't find it cardioprotective, either. In other words, if this had been the WHI, it wouldn't have been stopped,” explained Dr. Barnhart, who is with the department of obstetrics and gynecology at the university.

Women were selected for the retrospective study if their demographics matched those of the WHI cohort. They either took 0.625 mg daily of conjugated estrogen and 150 mcg of norgestrel on days 17–28 per cycle, or they served as controls.

In the GPRD, the adjusted hazard ratio for myocardial infarction was 0.95 (0.78–1.16) for older women and 0.91 (0.69–1.20) for younger women.

By contrast, in the WHI, the hazard ratio for nonfatal MI was 1.28 (0.96–1.70); for coronary heart disease deaths (including fatal MI), it was 1.10 (0.65–1.89) (N. Engl. J. Med. 2003;349:523–34).

Death from all causes was significantly lower in the older GPRD subjects taking HT than in control subjects, with a hazard ratio of 0.75 (0.65–0.86). It also was lower among the younger women taking hormones vs. younger controls (hazard ratio 0.76 [0.63–0.91]). In the WHI, overall mortality was not affected by HT.

In an interview at the meeting, Dr. Barnhart called the reduced mortality finding “mildly surprising.”

“Part of me wants to say that death is really the only thing that matters,” he said.

“You could have a stroke or have breast cancer or you could be protected from colorectal cancer, but really it matters what happened to you, and it looked like there was a lower death rate. I don't know why.”

Analyses are underway to determine whether missing data may help to account for the mortality findings that differed from the WHI results.

In some respects, the GPRD study closely paralleled WHI conclusions. For example, elevated rates of stroke were seen in both studies, with hazard ratios of 1.23 in the GPRD and 1.41 in WHI results that were reported after the trial was discontinued early (JAMA 2002;288:321–33). Similarly, breast cancer was elevated among hormone users in both trials, with hazard ratios of 1.67 and 1.26 in the GPRD and WHI, respectively.

Venous thromboembolic events were elevated in hormone users in both trials, with hazard ratios of 1.55 in the GPRD and 2.22 in the WHI, while the risk of colorectal cancer was diminished (with a hazard ratio of 0.56 in the GPRD and 0.63 in the WHI).

The hazard ratios for hip fractures were similar in both studies, 0.82 in the GPRD and 0.66 in the WHI.

Dr. Barnhart said the sheer size and scope of the GPRD, plus the meticulous inclusion criteria, served to overcome bias often associated with observational studies as opposed to randomized trials such as the WHI.

“If you have a relatively small observational study, then you've got the possibility of selecting women that might be healthier. But when you look at a large number of women, a cross section of the population—this is close to 10% of the population—you're much more likely to see what's actually happening in real life.”