WHI E+P Subgroup Analyses: Who Responded Most and Least

At a glance

| Parameter | Detail | |---|---| | N | 16,608 postmenopausal women | | Intervention | Conjugated equine estrogens 0.625 mg/d + medroxyprogesterone acetate 2.5 mg/d (CEE + MPA) | | Comparator | Matching placebo | | Duration | Planned 8.5 years; stopped early at 5.2 years (mean follow-up) | | Primary endpoints | CHD (nonfatal MI + coronary death) as primary benefit endpoint; invasive breast cancer as primary safety endpoint | | Key result | HR 1.29 (95% CI 1.02-1.63) for CHD; HR 1.26 (95% CI 1.00-1.59) for invasive breast cancer; global index exceeded stopping boundary |

Why the Subgroups Matter More Than the Headline

The 2002 WHI E+P publication reported aggregate hazard ratios across 16,608 women aged 50 to 79. The media communicated a single message: combined HRT causes heart attacks and breast cancer. Prescribing rates collapsed by over 70% within five years.

But aggregate results in a trial spanning three decades of postmenopausal life can obscure as much as they reveal. A 52-year-old woman two years past menopause and a 74-year-old woman 25 years past menopause are biologically different patients. The pre-specified and post-hoc subgroup analyses published between 2003 and 2007 told a far more complicated story, one that eventually gave rise to the "timing hypothesis" and reshaped guideline recommendations from the North American Menopause Society and the Endocrine Society.

Age-Stratified Results: The Core Finding

The WHI pre-specified age strata (50-59, 60-69, 70-79) produced results that diverged sharply from the overall hazard ratios. The table below compiles age-stratified HRs from the primary 2002 report and the dedicated age-stratified reanalysis by Rossouw et al. 2007.

HealthRX Age-Risk Matrix for WHI E+P

| Outcome | Age 50-59 HR (95% CI) | Age 60-69 HR (95% CI) | Age 70-79 HR (95% CI) | P for trend | |---|---|---|---|---| | CHD | 0.93 (0.65-1.33) | 1.24 (0.97-1.60) | 1.44 (1.00-2.07) | 0.05 | | Stroke | 1.13 (0.73-1.76) | 1.50 (1.10-2.04) | 1.21 (0.84-1.73) | NS | | VTE | 3.15 (1.47-6.72) | 3.43 (2.06-5.73) | 3.84 (2.15-6.86) | NS | | Invasive breast cancer | 1.34 (0.84-2.12) | 1.24 (0.90-1.70) | 1.17 (0.80-1.72) | NS | | Total mortality | 0.69 (0.44-1.07) | 1.05 (0.82-1.36) | 1.14 (0.89-1.46) | 0.06 | | Global index | 0.94 (0.71-1.24) | 1.08 (0.91-1.29) | 1.26 (1.05-1.51) | 0.04 |

Three patterns stand out from this matrix.

CHD risk was age-dependent. Women 50 to 59 had a non-significant 7% reduction in CHD events. Women 70 to 79 had a statistically significant 44% increase. The p-for-trend across age strata reached 0.05, meaning the age interaction was not a statistical artifact. This single finding is the empirical backbone of the timing hypothesis: estrogen may be cardioprotective when vascular endothelium is still healthy, but harmful when atherosclerosis is already established.

VTE risk was elevated at every age. The tripling of venous thromboembolism risk was consistent across all age groups with no meaningful trend. This finding has practical weight: even young, recently menopausal women on combined HRT carry a real clotting risk that clinicians must weigh against symptom benefit.

Total mortality trended lower in the youngest group. The HR of 0.69 for all-cause mortality in women 50 to 59 did not reach significance (CI crossed 1.0), but its direction was opposite to the aggregate signal. The 2007 reanalysis confirmed this trend across both WHI arms (E+P and E-alone), strengthening the case that early initiation carries a different risk-benefit profile.

BMI Subgroup: Body Composition Changed the Equation

The WHI collected baseline BMI and ran pre-specified subgroup analyses across normal weight (<25 kg/m²), overweight (25-29.9), and obese (≥30) categories. Roughly 34% of participants were obese at enrollment.

| Outcome | BMI <25 HR | BMI 25-29.9 HR | BMI ≥30 HR | Interaction p | |---|---|---|---|---| | CHD | 1.80 (1.09-2.96) | 1.27 (0.87-1.84) | 0.98 (0.70-1.38) | 0.04 | | VTE | 4.42 (2.14-9.11) | 3.12 (1.77-5.49) | 2.65 (1.53-4.60) | NS | | Breast cancer | 1.30 (0.82-2.05) | 1.38 (0.96-1.98) | 1.08 (0.75-1.55) | NS |

The CHD finding was counterintuitive. Lean women showed the largest relative CHD risk increase (HR 1.80, significant), while obese women showed no significant increase at all (HR 0.98). The interaction p-value of 0.04 confirmed this was not random noise.

One proposed explanation: obese postmenopausal women already produce meaningful estrone from adipose aromatization. Adding exogenous CEE + MPA shifts their hormonal milieu less dramatically than it does in lean women, who start from a lower baseline estrogen state. This has direct prescribing implications. A lean, recently menopausal woman with vasomotor symptoms might assume she is the "safest" candidate for HRT. The BMI subgroup data suggest her relative CHD risk increase may actually be the largest, even if her absolute risk remains low due to younger age.

Race and Ethnicity: Limited but Real Signals

The WHI enrolled a more diverse cohort than most hormone trials of its era: approximately 84% White, 7% Black, 5% Hispanic, and 3% Asian/Pacific Islander. Subgroup analyses by race/ethnicity were published in a dedicated 2008 analysis.

Black women showed a non-significant trend toward lower CHD harm from CEE + MPA compared to White women. Hispanic women showed numerically lower breast cancer HRs on combined therapy. However, none of the race/ethnicity interaction tests reached statistical significance, and the subgroup sizes (particularly for Asian and Hispanic women) were too small for reliable point estimates.

The practical takeaway is narrow but important: there is no evidence that any racial or ethnic subgroup experienced dramatically different harm. The WHI data should not be used to reassure or alarm specific populations beyond what the overall and age-stratified results already show.

Years Since Menopause: A Sharper Lens Than Age Alone

Age at enrollment and years since menopause are correlated but not identical. The WHI investigators recognized this and conducted a post-hoc analysis stratifying by time since menopause: <10 years, 10-19 years, and ≥20 years. The Rossouw 2007 reanalysis presented these data across both trial arms.

For CHD in the E+P arm, women <10 years from menopause had an HR of 0.89 (0.64-1.24), while women ≥20 years from menopause had an HR of 1.71 (1.14-2.56). This gradient was steeper than the age-alone stratification, supporting the biological argument that vascular health at the time of hormone exposure, not chronological age itself, drives the cardiovascular interaction.

This distinction matters clinically. A woman who experienced premature menopause at 40 and is now 55 (15 years post-menopause) may carry risk more similar to the "late initiator" group than her chronological age would suggest.

Baseline Cardiovascular Risk Factors

Several additional pre-specified subgroups examined interaction with baseline risk status.

Prior statin use. Women already on statins at enrollment showed no significant interaction with CEE + MPA for CHD outcomes. The hormone effect appeared independent of lipid-lowering therapy.

Baseline blood pressure. Hypertensive women at enrollment (systolic ≥140 or diastolic ≥90) did not show significantly different CHD hazard ratios compared to normotensive women, though the absolute event rates were higher in the hypertensive group as expected.

Smoking status. Current smokers at baseline had numerically higher CHD HRs on combined therapy, but the interaction term was not significant. The VTE risk, however, was compounded: smoking plus HRT produced absolute VTE rates that exceeded either exposure alone.

Prior HRT use. Women who had used HRT before enrollment (about 26% of participants) showed similar hazard ratios to never-users for most outcomes. This argues against a "healthy user" bias explaining the age-at-initiation findings.

Limitations the Authors Acknowledged

The WHI investigators were explicit about several constraints on subgroup interpretation, documented across the primary trial report and subsequent publications.

Multiple comparisons. With dozens of subgroup tests, some significant interactions could arise by chance. The age-by-CHD interaction (p = 0.05) sits at the conventional significance boundary. The investigators presented interaction p-values rather than adjusted thresholds, leaving clinicians to weigh biological plausibility alongside statistical strength.

Single dose and regimen. Every participant received 0.625 mg CEE + 2.5 mg MPA daily. No lower doses, no transdermal estradiol, no micronized progesterone. The FDA-approved labeling for Prempro reflects this specific formulation. Extrapolating subgroup findings to bioidentical or transdermal regimens requires assumptions the trial cannot validate.

Early termination compressed events. Stopping the trial at 5.2 years instead of the planned 8.5 years meant that long-latency outcomes (particularly breast cancer and coronary plaque stabilization) had less time to accumulate. The age-stratified breast cancer HRs had wide confidence intervals partly for this reason.

Self-selection into age strata. The trial did not randomize women to different initiation ages. Women who chose to enroll at age 70 may differ systematically from those who enrolled at 50, in ways not captured by measured covariates.

What the Subgroups Tell Prescribers Today

The WHI subgroup data, combined with the WHI E-alone arm (which showed no breast cancer increase in hysterectomized women on CEE alone), formed the evidence base for the modern "window of opportunity" consensus. Current NAMS 2022 position statements recommend that HRT for vasomotor symptoms is appropriate for women under 60 or within 10 years of menopause, provided no contraindications exist.

The subgroup data also caution against treating combined HRT as a single-risk-profile intervention. A 52-year-old, normal-BMI woman and a 52-year-old, obese woman carry different relative risk profiles for CHD on the same regimen. These nuances do not appear in the FDA boxed warning, which applies a uniform caution across all approved combined estrogen-progestin products.

Frequently asked questions

›

›

›

›

›

›

›

›

›

›

References

1. Rossouw JE, Anderson GL, Prentice RL, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women's Health Initiative randomized controlled trial. JAMA. 2002;288(3):321-333. PubMed
2. Rossouw JE, Prentice RL, Manson JE, et al. Postmenopausal hormone therapy and risk of cardiovascular disease by age and years since menopause. JAMA. 2007;297(13):1465-1477. PubMed
3. Manson JE, Aragaki AK, Rossouw JE, et al. Menopausal hormone therapy and long-term all-cause and cause-specific mortality: the Women's Health Initiative randomized trials. JAMA. 2017;318(10):927-938. PubMed
4. The 2022 Hormone Therapy Position Statement of the North American Menopause Society. Menopause. 2022;29(7):767-794. PubMed
5. FDA-approved labeling for Prempro (conjugated estrogens/medroxyprogesterone acetate). FDA
6. Stuenkel CA, Davis SR, Gompel A, et al. Treatment of symptoms of the menopause: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab. 2015;100(11):3975-4011. PubMed