Example• NHS study

– Exposed based on preferences/symptoms, younger population of women, shorter term hormone use

– Analysis of women taking hormones (more like per protocol)– Observe hormone use associated with decreased heart disease

rates• WHI Trial

– Randomized exposure, older population of women, longer term hormone use

– Analysis based on treatment assigned by randomization (ITT analysis)

– Observe hormone use associated with increased heart disease rates

Summary• Other observational designs are best understood

as sampling from a cohort experience– Certain case-control designs sample from a cohort

moving through time

• Note: we will do a set of group discussion after we have completed all the observational designs

Extra slides (Women’s Health Initiative)

Example - WHI• Between 1993 and 1998 Women’s Health

Initiative (WHI) enrolled 161,809 postmenopausal women aged 50-79 in a set of clinical trials with factorial design– Low fat diet– Calcium and Vit D supplementation– Two trials of postmenopausal hormone use

• Involved 40 recruitment sites in the US

Example - WHI• Trial of combined estrogen and progestin

conducted to assess whether conferred cardioprotection in women with a uterus

• Estrogen and progestin are used together in women with a uterus in order to diminish the risk of endometrial cancer

Example - WHI• Heart disease selected as main outcome

because it is the major cause of morbidity and mortality among postmenopausal women, especially over the age of 65

• Hypothesis of cardioprotection based on– Supportive data on lipid levels in trials– Trials in non-human primates– Large body of observational studies suggesting 40-

50% reduced risk

Example - WHI• Hip fracture as secondary outcome• Breast cancer designated as primary adverse

outcome• Other secondary outcomes

– Other cardiovascular diseases– Other cancers including endometrial– Other fractures

Example - WHI• Women with a uterus randomized to either

estrogen plus progestin or placebo

Example - WHI• Inclusion criteria• Ages 50-79 included

– Would have simplified to restrict ages 55-69• 50-55 requires verification of postmenopausal status• 69-79 may have less control over food choice, ability to

participate in activities over 9 year follow-up (other trial components)

– However HRT relevant to early postmenopause period

– Older women will have higher rates of the key outcomes

Example - WHI

Example - WHI• Monitoring began in Fall 1997• Late 1999 small but consistent early adverse

effects in cardiovascular outcomes and the global index– Contacted participants to communicate increases in

cardiovascular outcomes but trial continued because balance of benefits and risks remained uncertain

Example - WHI• 2002 interim analyses found adverse

cardiovascular outcomes persisted, although still within monitoring boundaries

• However breast cancer had crossed the monitoring boundary and the global index supported overall harm

• Early stopping of the estrogen plus progestin component

• Unopposed estrogen arm continued

Example - WHI

Example - WHI• Hypothesized that divergent findings from

observational studies and randomized clinical trials may be at least partially attributable to– Differences in the populations studied regarding age,

years since menopause, underlying risk of CHD– Methodologic limitations of observational studies

• Note that trial did not address potential risks/benefits of short term estrogen plus progestin for menopausal symptom management

Example