Lessons from the Mammography Wars REPLY

Abstract

In this article, the authors argue that any intervention produces a continuous gradient of benefit versus harm. Subjective thresholds can vary, even if the data are agreed upon. Should we rely solely on expert panels to make arbitrary judgments, especially in the gray areas?

Full text

sounding b oa r d
The new england journal of medicine
n engl j med 363;11 nejm.org september 9, 2010
1076
Lessons from the Mammography Wars
Kerianne H. Quanstrum, M.D., and Rodney A. Hayward, M.D.
People of the same trade seldom meet to-
gether . . . [without] the conversation end-
[ing] in a conspiracy against the public.
— Adam Smith, The Wealth of Nations, 1776
1
The controversy was predictable.
Since 2002, annual mammograms had been
recommended for women 40 years of age or old-
er.
2
Suddenly, an independent, government-fund-
ed panel was suggesting that this schedule might
be too much — that less, in fact, might be better.
Advocates of breast-cancer screening, particu-
larly breast radiologists, immediately took action,
denounci ng the panel’s statement s as government
rationing, suggesting that the panel members had
ignored the medical evidence, and even implying
that the panel members were guilty of a callous
disregard for the life and well-being of women.
As one prominent breast radiologist put it, “Ba-
sically, [the panel] said nothing is good. Just wait
until it breaks through your skin. . . .”
3
Speci alt y
societies quickly issued countermanding guide-
lines.
4
In reality, this independent panel, the Preven-
tive Services Task Force, simply recommended t hat
routine screening mammography begin at the age
of 50 years, whereas women bet ween the ages of
40 and 49 years should make individual decisions
with their doctors as to whether their preferences
and risk factors indicate screening at an earlier
age. The panel also recommended that screening
mammograms be performed every other year,
which they suggested would reduce the harms of
mammography by nearly half while maintaining
most of the benef its provided by annual imag-
ing.
5
In short, the panel concluded that we had
previously overestimated the value of mammog-
raphy: that mammography is good, but not that
good; that it is necessary for many women, but
not all; and that it should be performed at some
frequency, but perhaps not every year, for every
woman.
Behind the panel’s conclusions regarding mam-
mography lurks an unwelcome reality that our
profession has often failed to acknowledge. Every
medical intervention — no matter how benefi-
cial for some patients — will provide continu-
ously diminishing returns as the threshold for in-
tervention is lowered. Mammography is just one
case in point. For women bet ween the ages of 40
and 49 years, the false positive rate is quite high,
and the expected benefits are quite low: more
than 1900 women would need to be invited for
screening mammography in order to prevent just
one death from breast cancer during 11 years of
follow-up, at the direct cost of more than 20,000
visits for breast imaging and approximately 2000
false positive mammograms. Conversely, for
women between the ages of 60 and 69 years, fewer
than 400 women would need to be invited for
screening in order to prevent one breast-cancer
death during 13 years of follow-up, while accruing
approximately 5000 visits and 400 false positive
mammograms.
6
In short, as the risk of breast
cancer increases, the benef its of mammography
increase, whereas the relative harms become pro-
gressively less signif icant.
More generally, the net benefit of all medical
treatments is a continuous function of three fac-
tors: the risk of morbidity or mortalit y if untreat-
ed (Risk
NoRx
), the treatment’s relative risk reduc-
tion (RRR
Rx
), and the treatment’s risk of harm
(Harms
Rx
):
Net Benefit = (RiskNoRx × RRRRx) – (HarmsRx).
As the risk of no treatment (Risk
NoRx
) decreas-
es, the net benefit of treatment will decrease, even
if the treatment’s relative benef it (RRR
Rx
) remains
constant. Indeed, for many interventions, if the
risk of no treatment is low enough (e.g., if we low-
er the threshold for treatment too far or if a pa-
tient’s life expectancy is relatively limited for other
reasons), then the side effects and risks of treat-
ment will dominate, and the treatment will result
The New England Journal of Medicine
Downloaded from www.nejm.org by ARCHIE BLEYER on September 13, 2010. For personal use only. No other uses without permission.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.

n engl j med 363;11 nejm.org september 9, 2010
1077
sounding board
in net harm.
7-9
Since the risk of no treatment var-
ies dramatically among patients for almost every
disease or condition, even a highly effective inter-
vention will show a gradient of net benefit in a
given population.
Despite this continuous gradient of treatment
benef it versus harm, medical decision making is
necessarily discrete. In the case of any given pa-
tient, we must choose to treat or not treat, to
screen or not screen. In an effort to help us make
these choices, our profession is constantly trying
to elucidate clear thresholds for intervention, such
as the level of glycated hemoglobin or low-densi-
ty lipoprotein cholesterol, age, or standard time
intervals. What we often do not remember is that
these thresholds — for example, an age of 40 ver-
sus 50 years or annual versus biennial routine
mammography — are to some degree subjective
and arbitrar y. After all, scientif ic evidence can only
help us describe the continuum of benefit versus
harm. The assessment of whether the benefit is
great enough to warrant the risk of harm — i.e.,
the decision of where the threshold for interven-
tion should lie — is necessarily a value judgment.
When either side in the mammography wars
claims that the evidence suggests that women
should or should not undergo routine mammog-
raphy starting at the age of 40 years, they are de-
ceiving themselves and the public about what the
evidence can tell us. They are really just making
different value judgments about where to set the
threshold.
Who is right? Who should be making these
judgments?
The obvious answer might seem to be “the in-
dividual patient and her doctor.” But it would be
folly to suggest that every medical decision ought
to be made anew for each patient, with no stan-
dard of care in any case, no guidelines, and no
professional norms. Instead, our profession needs
to start distinguishing between choices that are
clear-cut and those that require individualized
decision making. To this end, for most interven-
tions, rather than seeking a single, universal
threshold for intervention (
Fig. 1A
), we should be
arguing over a minimum of two distinct thresh-
olds: one above which benefit clearly outweighs
the risk of harm, in which c ase clinicians should
recommend a treatment; and one below which
concern about harm clearly dominates, in which
case clinicians should recommend against that
treatment. Between these two thresholds lies a
gray area of indeterminate net benefit, in which
clinicians should defer to an individual patient’s
preferences — including, for example, a woman’s
emotional response to her risk of breast cancer
— in choosing whether to intervene (
Fig. 1B
).
It is just such a gray area into which women
in their 40s are assigned by the new mammog-
raphy guidelines. Of note, there are quantitative
methods available that can assist clinicians in
guiding individual decision making even in these
gray areas.
6-10
In our profession’s apt pursuit of
more systematized care, however, we have gen-
erally preferred to ignore these gray areas alto-
gether. It is easier, after all, to simply lower the
threshold for inter vent ion — to recommend mam-
mography for all women 40 years of age or older
— than to rely on individual judgments about
which of these women actually warrant screening.
However, our current approach is more than
just a quest for uniformit y. When a given service
is successfully extended to more people with more
intensity, the profession providing that service
tends to grow in importance and profitability. In
the United States, where medical specialists often
enjoy an exalted status in the minds of the pub-
lic, if experts shout loudly that every woman 40
years of age or older must be screened annually
for breast cancer, then breast cancer must be im-
portant, screening must be a basic human right,
and doctors who provide this service must have
great value and authority.
But what if those experts are basing their rec-
ommendations on more than the interest of pa-
tients alone? In any other industry, we accept the
idea as natural that those providing a service or
BNew Model
ACurrent Model
Net benefitNet harm
Care
discouraged
Care recommended
Substantial
net benefitSmall or uncertain net benefit
Net harm
Care
discouraged
Care
recommended
Discretionary care
Figure 1. Creating a New Risk–Benefit Model that Allows for Individualized
Decision Making.
For most interventions, the current practice of seeking a single, universal
threshold for intervention (Panel A) might be replaced by a model that allows
for individualized decisions about whether to inter vene on the basis of per-
sonal risk factors and preferences (Panel B).
The New England Journal of Medicine
Downloaded from www.nejm.org by ARCHIE BLEYER on September 13, 2010. For personal use only. No other uses without permission.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.

The
new england journal
of
me dicine
n engl j med 363;11 nejm.org september 9, 2010
1078
product hold their own and their shareholders’
interests as a primary objective. Why have we
failed to acknowledge that the same phenome-
non occurs in health care? Although it is true that
individual medical providers care deeply about
their patients, the guild of health care profession-
als — including their specialty societies — has
a primary responsibility to promote its members’
interests. Now, self-interest is not in itself a bad
thing; indeed, it is a force for productivity and
eff iciency in a well-functioning market. But it is
a fool’s dream to expect the guild of any service
industry to harness its self-interest and to act ac-
cording to beneficence alone — to compete on
true value when the opportunit y to inf late per-
ceived value is readily available.
It is for this reason that some degree of mar-
ket regulation is necessary, such as truth-in-adver-
tising and antitrust laws. It is only in health care
that we have failed to recognize the need for anal-
ogous protections. It is only in health care, after
all, that the same group that provides a service
also tells us how valuable that service is and how
much of it we need, as when the Society of Breast
Imaging sets the recommendations for mammog-
ra phy.
4
If there is overutilization in health care,
we can be sure that it will continue unabated as
long as those with a vested interest are allowed to
win the public-relations wars by shouting about
“rationing” or “death panels” whenever anyone
suggests that more health care, in fact, may not be
bet ter.
It is time for a change. We must acknowledge
that just as in any other profession or industry,
self-interest is unavoidably at work in health care.
Rather than even acknowledging practice guide-
lines offered by vested experts, we ought to bor-
row from the wisdom of sound governance and
implement a system of checks and balances when
it comes to the interpretation and application of
medical evidence. At the same time, we need to
recognize that these two tasks are distinct. Al-
though the interpretation of medical evidence is
(or ought to be) a scientif ic exercise, the applica-
tion of that evidence, as in guideline formation,
is ultimately a social exercise.
Decisions regarding practice guidelines can,
and certainly should, be informed by evidence. But
they will always require value judgments regard-
ing how much evidence is suff icient to dictate
care, for example, or whether and to what degree
costs should be considered. By separating the pro-
cesses of evidence review and guideline format ion,
fair disagreements about the quality or substance
of the evidence can occur separately from, and be-
fore, disagreements about the implications for
clinical care.
Ideally, we ought to have a system in which
independent panels of generalists, with expertise
in the methods of evidence review and synthesis,
are responsible for objectively synthesizing the
medical evidence around a given question or pro-
cess of care. These independent panels could then
seek input from the relevant clinician groups re-
garding their views concerning the evidence and
where they feel the thresholds for recommended
care versus individualized decision making should
reside. To facilitate impartiality and political vi-
ability, a public–private alliance might be best,
with funding and representation for the indepen-
dent panels coming from government, private
foundations, and provider and payer groups.
Furthermore, unlike one-time or occasional
panels, t h is process of evidence review a nd guid e-
line formation ought to be adequately funded to
allow for regular updates as new evidence be-
comes available. Recent proposals to increase
spending on comparative-effectiveness research
are certainly laudable, but it is unrealistic to think
that an investment in research alone will have a
sizable effect on the practice of medicine without
a concomitant investment in a credible process
for vetting medical evidence and clinical care
guidelines.
The Preventive Services Task Force approaches
the format that we propose here, since the panel
is composed of expert generalists. This panel,
however, tends to interpret t he evidence and write
the recommendations as a single process, creating
the appearance, and perhaps the realit y, of al-
lowing too little input from vested interests and
often conf lating disagreements about the evidence
with disagreements about the recommendations.
Furthermore, as seen in the mammography wars,
the panel’s perceived failure to seek sufficient in-
put from specialty groups was a prominent, and
seemingly effective, argument against the guide-
lines.
11
Even the perception that the process of
guideline formation is closed or does not consid-
er specialists’ opinions can make it easy for those
arguing that more care is better to prey on the
public’s legitimate concern that government and
insurers are out to deny them lifesaving care.
As a profession, we have the potential to play
The New England Journal of Medicine
Downloaded from www.nejm.org by ARCHIE BLEYER on September 13, 2010. For personal use only. No other uses without permission.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.

n engl j med 363;11 nejm.org september 9, 2010
1079
sounding board
a very real role in improving our health care sys-
tem. We can choose t o acknowledge the gray a reas
of medicine and insist that they be ref lected in
clinical-practice guidelines and in performance
measures. And we can work to prevent vested in-
terests from being granted the loudest voices in
hea lt h care — even when those voices bla zon f rom
our own specialty — by granting credence to
groups such as the Preventive Services Task Force
that seek to formulate evidence-based guidelines
in an objective way.
Or we can, instead, conduct our own version
of the mammography wars when a prudent ap-
plication of the evidence threatens the profit-
ability and stature of our own specialty.
The opinions expressed in this article are those of the authors
and do not necessarily represent those of the Robert Wood
Johnson Foundation or the University of Michigan.
Disclosure forms provided by the authors are available with
the full text of this article at NEJM.org.
From the Robert Wood Johnson Foundation Clinical Scholars
Program (K.H.Q., R.A.H.); the Departments of Surgery (K.H.Q.)
and Internal Medicine (R.A.H.), University of Michigan; and the
Department of Veterans Affairs (VA), VA Health Services Re-
search and Development Center of Excellence, VA Ann Arbor
Healthcare System (R. A.H.) — all in Ann Arbor, MI.
Smith A. The wealth of nat ions, part one. New York: P.F.
1.
Collier & Son, 1911:207.
U.S. Preventive Services Task Force. Screening for breast
2.
cancer: recommendations and rationale. A nn Intern Med 2002;
137:344-6.
Bryner J. New mammogram guidelines issued . . . aga in. 3.
Januar y 4, 2010. (Accessed August 19, 2020, at http://ww w
.livescience.com/health/100104-breast-cancer-screening.html.)
Lee CH, Der shaw DD, Kopans D, et al . Breast cancer scr een-4.
ing wit h imag ing: recommendat ions from the Societ y of Breast
Imagi ng and the ACR on the use of mammography, breast MRI,
breast ult rasou nd, and other technolog ies for the detection of
cli nic al ly oc cult bre ast c ancer. J A m Col l Radiol 2010;7:18-27.
U.S. Preventive Services Task Force. Screening for breast 5.
cancer: U.S. Preventive Serv ices Task Force recommendation
stateme nt. A nn I nte rn Me d 200 9;151:716-26.
Nelson HD, Tyne K, Naik A, et al. Screening for breast can-6.
cer: an update for the U.S. Prevent ive Services Task Force. A nn
Intern Med 20 09;151:727-37.
Hayward RA, Kent DM, Vijan S, Hofer TP. Reporti ng clinical 7.
tr ia l resu lt s to in form prov iders , pa yer s, a nd consum ers. He al th
Aff (Mi llwood) 2005;24:1571-81.
Hay wa rd R A, Kr umholz HM, Zul ma n DM, Timbie J W, Vijan 8.
S. Optimizing statin treatment for primary prevention of coro-
nar y a rtery diseas e. Ann Inter n Med 2010;152:69-77.
Zulma n DM, Vijan S, Omenn GS, Haywa rd RA. The relative 9.
merits of population-based a nd t argeted prevent ion strat egies.
Milbank Q 2008;86:557-80.
Vickers AJ, Croni n AM. Tradit ional st atistical methods for 10.
evaluating prediction models are uninformative as to clinical
value: t oward s a decision analyt ic f ra me work. Se mi n Onc ol 2010;
37:31- 8.
Walker EP. USPSTF members defend ne w gu idelines before 11.
Congr ess. (Acc es sed Augu st 1 9, 2010, at ht tp://w ww.medp aget od ay
.com/HematologyOncology/BreastCancer/17272.)
Copyri ght © 2010 Mass achusetts M edical Soc iety.
electronic
a c c e s s
t o
t h e
j o u r n a l
’
s
cumulative
i n d e x
At the Journal’s site on the World Wide Web (NEJM.org),
you can search an index of all articles published since January 1975
(abstracts 1975–1992, full text 1993–present). You can search by author,
key word, title, type of article, and date. The results will include the citations
for the articles plus links to the full text of articles published since 1993.
For nonsubscribers, time-limited access to single articles and 24-hour site
access can also be ordered for a fee through the Internet (NEJM.org).
The New England Journal of Medicine
Downloaded from www.nejm.org by ARCHIE BLEYER on September 13, 2010. For personal use only. No other uses without permission.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.