Originally published in HCPLive.com
by Don S. Dizon, MD, FACP
Recently, two groups released new guidelines that may affect breast cancer and cervical screening in women.
The US Preventive Services Task Force (USPSTF) recommended against screening women in their 40s and to screen women every other year starting at age 50. The USPSTF left room for individualized screening, particularly in the presence of risk factors such as family history. Still, it is a departure from previous recommendations (and current standard) for annual screening of women starting in their 40s.
For cervical cytology screening, the American College of Obstetricians and Gynecologists (ACOG) revised its pap smear guidelines, calling for women to receive them every two years between the ages of 21 and 30. The revised recommendations say that women age 30 and older who receive three consecutively negative test results may be screened once every three years. This, too, is a departure, as previous guidelines recommended annual cervical cytology screening beginning three years after the onset of sexual activity. Both revised guidelines have ignited a firestorm of controversy, though the response to the cervical cancer screening guidelines has been less severe.
I do not believe the controversy stems from the actual recommendations being made. In both guidelines, a panel of experts reviewed the latest information before recommending changes. What is interesting is how the data were actually used. In the mammography guidelines, the data were used to generate mathematical models of benefits and harms for screening. Using these models, the reviewers determined that 1,900 women between the ages of 40 and 49 would be invited to be screened to save one woman from breast cancer. Still, the benefit was statistically significant, as noted in the relative risk of 0.85 (CI, 0.75 to 0.96). The fact that relative risk does not cross 1 makes it so. So, here we have a statistically significant finding, which the USPSTF takes as non-clinically significant. But then, how much is a life worth?
The ACOG recommendations were data-driven as well, but pooled from the evidence. No modeling was performed. Instead, the biology and natural history of HPV and of cervical cancer were taken in to account, coupled with the long-term consequences of intervening early in a woman’s life. The fact that invasive cervical cancer represents a step-wise progression from abnormal changes in the cervix (which all can be picked up with cervical screening) makes cervical cancer a rare disease in women under age 21 and explains why three consecutively negative pap smears does not necessitate continued annual testing.
Still, there is another difference. ACOG recommendations represent those of the group that is doing the pap smears, interpreting them for their patients, acting on the abnormal results, and then seeing the consequences of their intervention. The USPSTF cannot be clearly deemed to be the same. These are not oncologists, surgeons, and/or breast imagers. Instead, the USPSTF is composed of medical experts in primary care. They are not breast specialists by any measure, so their recommendations of what constitutes “unreasonable risks” are bound to be controversial, particularly when the risk women are most alarmed about is “death from breast cancer.”
Don S. Dizon is an oncologist who blogs at The Women’s Cancer Blog.
Submit a guest post and be heard.
Similar Posts:
- Informed consent is missing from Pap smears and cervical cancer screening
- When women should have their first Pap smear; the new cervical cancer screening guidelines
- Will patients accept the new, evidence-based, breast cancer screening guidelines?
{ 13 comments }
Interesting that the uspstf’s meta-analysis found a clinically significant difference in the 40-49 age group but the studies that they used in the analsis overwhelmingly did not. Here are some examples:
The Age trial (one of the best):(breast ca mortality) rr 0.83 (0.66-1.04)
overall mortality rr 0.97 (0.89-1.04)
Stockholm (Breast cancer Death) rr 1.08 (0.54-2.17)
Gothenburg (breats ca death)rr 0.69 (0.45-1.05)
CNBSS-1 (breast ca death) rr 1.06 (0.80-1.40)
How does this make sense?
Still, the benefit was statistically significant, as noted in the relative risk of 0.85 (CI, 0.75 to 0.96). The fact that relative risk does not cross 1 makes it so. So, here we have a statistically significant finding, which the USPSTF takes as non-clinically significant. But then, how much is a life worth?
What about the lives of healthy women who are harmed by overdiagnosis? How much are they worth? This is an unnecessarily inflammatory question that implies that the issue is only the cost and not actual HEALTHY women harmed by screening. There is a reason why oncologists, radiologists and the media have long been afraid to let women know the real statistics about both an individual probability of benefit and individual probability of harm. You are afraid that this information would turn the women off screening. If you think that this probability of having one’s life saved is high enough and well worth the harms, why not give the women complete and honest information about the probability of benefit, overdiagnosis, and the high probability of having at least one false positive and let them decide for themselves?
I am yet to see an oncologist or any doctor for that matter to honestly talk about overdiagnosis in mammography on TV. As an oncologist, would you care to answer the question as to why your collegues deem it necessary to lie to women?
As to oncologists on the panel – can oncologists estimate the extent of overdiagnosis or quantify the harms? Screening benefit may apply to oncology, but screening harms are mostly outside of their area, so they aren’t exactly objective.
This post is oddly reasoned.
Why are The USPTF “controversial” because they are not oncologists, OBs? Au contraire, recs from professional organizations are always susepct becayse their members profit from more screening.
This comment is strange: “The fact that relative risk does not cross 1 makes it so. So, here we have a statistically significant finding, which the USPSTF takes as non-clinically significant. But then, how much is a life worth?”
CB analysis does not put a value on life. That crude understanding is why the USPTF’s recs were so controversial–and doctors who don’t understand cb shouldn’t write about it. CB compares the cost of testing compared to the increased RISK of death. That is not putting a value on life. Everytime we get into a car to avoid walking or carrying groceries home we make the same calculation (increased risk of car accident death vs. convenience.)
Agree with posts 2 and 3. I prefer epidemiologists to oncologists on this issue.
So oncologists should not be trusted to give opinions regarding cancer screening and treatment because they treat cancer for a living and might profit from the results. Ok fair enough. I assume this also applies to other specialties as well. Maybe next time I think I might be having an MI I’ll get a second opinion from a neurologist to make sure the cardiologist isn’t trying to line his pockets by cathing me.
Does anyone else find this line of reasoning absurd? The idea that experts from a field should be excluded from practice guideline decisions is misguided. The only other organization I can think of that adopts this line of reasoning is Congress.
No one says “experts” from a field should be excluded from anything. However, medical specialty organizations are not “experts.” They are political entities that respond to their members’ needs and desires. Not always and not perfectly–here ACOG recommends less screening. But, the profit motive is always present in doctors’ advice–and to fail to recognize it foolish and naive.
Further, oncologists are not statisticians. They are not equipped nor trained to see the big picture. Of course, the big picture is only part of deciding a course of treatment, i.e., individual difference can be determinative, not population averages. However, for recommending global guidelines the big picture is most important–and more important than any practitioner’s puny perspective.
Frank Coburn – the experts in the field of evaluating data from studies or designing studies and models are not oncologists but epidemiologists i.e. people with expertise in biostatistics.
If you have an MI, you’d go to cardiologist, and the cardiologist will treat you and maybe prescribe you a drug to prevent another MI. However, it’s not cardiologists but statisticians who evaluated data from studies of this drug and determined that it worked. Since the cardiologist only sees individual cases, he or she cannot determine if the drug works or if his patients’ didn’t have another heart attack because of the drug but for some other reason such as for example lifestyle changes, luck or something else. This determination can be made by experts whose job is to design the studies in a way that would eliminate confounding factors, but also to evaluate the data, correlate the information, adjust for biases. Cardiologists aren’t trained to do it, it’s not their job.
Similarly with studies of screening. Oncologists see individual cases. They see a early case of cancer and they see that the person survives longer. Is it because of early detection or because of lead-time bias? The oncologist wouldn’t know that since this cannot be determined by simply treating individual cases, but an oncologist would love to think that it was because of early detection. Frankly, judging from some of the oncologists’ opinions, some of them cannot even spell lead-time bias… Similarly, is a particular cancer cured because it is detected early or was it detected early because it was slow growing or even not growing at all? Not only oncologists cannot determine it based on individual cases they see, they aren’t exactly objective. Nobody wants to think they harmed someone by treating them unnecessarily, it is much more pleasant to believe they cured the person. But even with all the best intention, only by evaluating large amount of data, analyzing the data for biases, correlating the data can these things be determined. Oncologists aren’t experts in determining if screening save lives or just detects cancers or some combination of this. It’s simply not their job.
“..cured because it is detected early or was it detected early because it was slow growing or even not growing at all?”
A Diora, that is length time bias not lead time bias in a screening exam. Anybody here actually read the JCO? Almost, every paper is full of stats and has statistician involvement. Spend a little more time reading before insulting.
I didn’t say this was lead time bias. These were different sentences and different examples.
Here is what I said:
“They see a early case of cancer and they see that the person survives longer. Is it because of early detection or because of lead-time bias? …”
This is what I said. The length of survival after diagnosis is affected by lead-time bias. The sentence you mention came later and was clearly a different thought. Maybe it should’ve been a different paragraph.
Of course, taking two different sentences out of context and deliberately misenterpreting them is one way to discredit someone you talk with…
Also, about the phrase you found insulting. Many a paper by an oncologist, especially when talking to the general public, talks about 5-year survival in the same context as screening. Sometimes it even happens in the research papers. If I have time, I could probably find zillion examples of this. This is what I had in mind when I wrote about oncologists and lead-time bias. Maybe I shouldn’t have said that – I normally try to be nice, but I heard doctors’ talk about how early detection increases 5-year survival so many times, and it makes me angry. It is one of my pet peeves.
Still, it was absolutly clear from my post that the sentence lead-time bias referred to the example before it and not the one that followed. Honestly, did you really miss what I mean or did your (mis)interpretation of my words was deliberate?
“There are three kinds of lies: Lies, Damned Lies, and Statistics.”
Benjamin Disraeli
Uh, that would Mark Twain.
Actually, Twain was quoting Disraeli (though it’s uncertain whether he actually coined the phrase).
From Wikepedia (yes, it does serve some purpose):
Twain popularized the saying in “Chapters from My Autobiography”, published in the North American Review, No. DCXVIII., July 5, 1907. “Figures often beguile me,” he wrote, “particularly when I have the arranging of them myself; in which case the remark attributed to Disraeli would often apply with justice and force: ‘There are three kinds of lies: lies, damned lies, and statistics.’”[1]
Comments on this entry are closed.