FDA or by industry. However, only 10 companies
(30%) frequently or usually collected data on men-
strual cycle; 56% replied that the FDA at some time
or other had requested theinclusionof women in
trials. When women of childbearing potential were
included in protocol proposals, 21% of the respon-
dents said that the FDA never disagreed, but 79%
had experience of some FDA reviewers at one time
or another excluding women of childbearing
potential.When excluded, this was usually in the
phase I and phase II trials, 58% and 45%, respec-
tively, correspondents reported.
Although this survey was qualitative rather than
quantitative, the results should not be dismissed
lightly; because the survey was confidential, no
respondents or their firms were exposed to open
criticism. Because of their experience and senior
positions, respondents had reviewed many differ-
ent drugs and NDA applications. The survey
replies were, therefore, likely to be reliable and
provide a good approximation of the then-current
industry gender practices and the frequency of
clinically meaningful differences.
When gender differences in safety or efficacy
were found to be clinically significant, most
respondent companies (94%) opted to put the
data in the product label, thePhysicians’ Desk
Referenceand the product literature (72%), and
to publish in the medical journals (69%). Presum-
ably, the two companies that did not amend their
labels acted thus because the products were only
intended for one-gender use. By December 1999,
there were 348 medicines in development for dis-
eases only in women or where women are dispro-
portionately affected (Holden, 2000). Not only
has industry stepped up its research efforts, but
many large firms have units devoted to women’s
healthcare.
Finally, correspondents were asked how fre-
quently gender differences were found; 73% said
‘occasionally’, 3% said ‘frequently’ and the rest
said ‘never’. Of those who saw differences, only
one-third found these differences to beclinically
significant 5% of the time, while 17% of respon-
dents said that significant differences occurred
10% of the time. This was more than expected,
and provides further justification for gender
testing.
16.7 Possible solutions
The author must stress that the opinions and the
suggestions that follow are personal, based on
30 years in industry, from phase I–IV study experi-
ence, with five large international pharmaceutical
firms.Women’s inclusion as drug research
subjectsWomen should be and, indeed, are included into
new drug and device development programs when
not specifically excluded due to male-only disease
or existing pregnancy. If it is predictable that a drug
or device will be used in women (though they may
not be the majority users), then a ‘reasonable num-
ber’ should be included in phase II and phase III
studies. If the disease occurs more frequently in
women, for example rheumatoid arthritis, then
women should be involved in phase I studies. The
reality is that of the many hundreds of drugs and
devices approved for use today, very few show
major gender-related differences in either side
effects or efficacy. Clearly, in the drug classes
that have been shown to demonstrate significant
gender clinical differences, ‘specific’ gender-
related studies should be included for investigation
drugs and devices. These could be similar to those
now undertaken in the elderly. First, a single-dose
study should be undertaken. If important differ-
ences are found compared to men, a multiple-
dose study ought to be undertaken, and then a
shorter duration efficacy and safety study in
women. Such studies can be conducted later, per-
haps concurrently with phase III of the develop-
ment program.
What do we mean by ‘a reasonable number’?
‘Reasonable’ is that number which would be
expected to show a significant genderclinicaldif-
ference if a real difference is present, and probably
will only apply to efficacy and adverse events 5%
or larger, because a difference in low-incidence
adverse events will not show up until the drug is
in the market. This would mean at least 300 women
exposed to the new drug. The number of patients
should be based on what is judged to be a clinically16.7 POSSIBLE SOLUTIONS 215