Page 8 - Mecklenburg_Medicine_October-2018

Feature
The Problem of Overdiagnosis
By Gordon Hull, Director, Center for Professional and Applied Ethics,
Associate Professor Philosophy and Public Policy, UNC Charlotte
Clinical tests always have to calibrate sensitivity (the ability
to avoid false negatives) and specificity (the ability to
avoid false positives). This is a well-known problem, and
is inherent to pretty much any process of looking for
something. The TSA has to calibrate its airport machinery to detect
explosives, but not to ring the alarm at deodorant. It’s been getting a lot
of attention in the context of mammography, as screening technology
gets more and more sensitive. Getting this right is hard, and of course,
is an ethical judgment at the end of the day. How much anxiety in false
positives is “worth it” for detecting an actual cancer? At what age is the
underlying risk great enough to warrant screening?
Today, I want to look at a different, perhaps even more difficult,
problem. It’s one that’s gotten less attention but is equally as serious.
“Overdiagnosis” happens when one treats as clinically relevant a tumor
that is not. More specifically, a current (June, 2018) paper by Louise
Davies et al.1 proposes defining it as “the detection of a (histologically
confirmed) cancer through screening that would not otherwise have
been diagnosed in a person’s lifetime had screening not been done.”
Lynette Reid, a bioethicist at Dalhousie University (and Ethics Center
speaker last year) frames her discussion of the problem with a pair of
seemingly contradictory data points. On the one hand, the evidence
says we are getting a lot better at breast cancer screening. Detection
rates are way up. On the other hand, mortality rates are barely budging.
The problem seems to be we are treating tumors that would never
be clinically significant. That the problem is overdiagnosis, and not
something else, is suggested by autopsy studies which show by the
time someone reaches old age, they almost certainly have a number
of indolent tumors in their body. This “disease reservoir” represents
a potentially serious overdiagnosis problem; Reid quotes research
indicating if all of the indolent tumors had been detected with
screening, then more than 99 percent of all thyroid cancers would
be overdiagnoses. Even now, perhaps nearly a third of breast cancers
are overdiagnoses. That said, a number of population-level variables
influence whether a cancer is likely an overdiagnosis, making such
estimates difficult. For example, Davies et al. point out that overdiagnosis
is less likely in a population with a lower life expectancy from time of
diagnosis, since a patient is more likely to die of other causes before a
slowly-growing cancer could become clinically relevant. So too, there
is a fundamental conceptual difficulty: whatever can be said at the
population level, it is impossible to know if a given cancer, once treated,
represented an overdiagnosis.
These conceptual difficulties turn into clinical ones. Davies et al. cite
a significant body of research indicating, in essence, that difficulties
in understanding and communicating screening risks and benefits of
screening generally, and of overdiagnosis specifically, make it hard to
reach clinical decisions guided by patients’ values and risk tolerances.
Even more fundamentally, Reid concludes by pointing out that the
statistics put us in a paradoxical situation, “The more overdiagnosed
8 | October 2018 • Mecklenburg Medicine
cancer we detect, the less predictive of clinically significant outcomes
it will be; at some point, its detection will confer no elevated risk of
mortality. At some point before we get that far … it will no longer make
sense to consider finding a risk factor (97).”
In other words, detecting a tumor may someday not actually predict disease
progression, or at least not well enough to guide clinical decision-making.
Importantly, though, overdiagnosis is real. Each overdiagnosis stands
for someone who will most likely undergo a physically-demanding and
potentially dangerous treatment regimen and (hopefully) emerge on the
other side with the socially-difficult label “cancer survivor.” In short,
and paradoxically, improvements in cancer screening are making ethical
choices about how to treat cancer harder, not easier.
1Defining, Estimating, and Communicating Overdiagnosis in
Cancer Screening. Louise Davies, MD, MS; Diana B. Petitti, MD,
MPH; Lynn Martin, PhD; Meghan Woo, ScD, ScM; Jennifer S. Lin,
MD, MCR. Annals of Internal Medicine, July 3, 2018.
NATIONAL HEALTH
& WELLNESS OBSERVANCES
OCTOBER 2018
Alzheimer’s Association Memory Walk n American Heart Walk
Eat Together, Eat Better Month n Children’s Health Month
Domestic Violence Awareness Month n Health Literacy Month n Healthy Babies Month
Healthy Lung Month n National Breast Cancer Awareness Month
National Bullying Prevention Month n National Chiropractic Month
National Dental Hygiene Month n National Depression and Mental Health Screening Month
National Family Sexuality Education Month n National Liver Awareness Month
National Physical Therapy Month n National Spina Bifida Awareness Month
National Sudden Infant Death Syndrome (SIDS) Awareness Month
Talk About Your Medicines Month n Vegetarian Awareness Month
October 1: Child Health Day
October 1-7 : National Mental Illness Awareness Week
October 8-12: National School Lunch Week
October 8-12: National Consultant Pharmacy Week
October 10: World Mental Health Day
October 10: Put the Brakes on Fatalities Day
October 14-20: International Infection Prevention Week
October 14-20: National Health Education Week
October 14-20: National Healthcare Quality Week
October 20: World Osteoporosis Day
October 21-27: National Respiratory Care Week
October 23-31: National Red Ribbon Celebration/National Plant the Promise Week
October 27 : Make a Difference Day
October 29: World Stroke Day