4 November 2015
Damned if you do. Damned if you don’t
No doctor can have missed the whirlwind of hype and concern about the MBS review into the billions wasted on overdiagnosis and overtreatment.
“Overdiagnosis is the biggest threat to the sustainability of the healthcare system,” says Dr Paul Glasziou, Professor of Evidence-Based Medicine at Bond University. “What it’s doing is moving resources away from the truly sick to the well.”
Of even greater concern, of course, is the risk for patients associated with needless tests and procedures, but this is difficult to quantify.
In the closing the provocative Four Corners program ‘Wasted,’ Kerri O’Brian told viewers that “Tonight’s story is not saying don’t have those tests and procedures. What it is saying is, ask your doctor the right questions before you proceed, like, ‘What is the evidence that they’re going to help me?'”
The problem is that the evidence is hard to understand even for health professionals and researchers, let alone the patient looking you in the eye for guidance.
Overdiagnosis occurs when people with no overt symptoms are diagnosed correctly as having some underlying condition or pathology. But if that condition would not lead to any symptoms within then the diagnosis can lead to the harmful consequences of overtreatment.
Overdiagnosis of this kind can be distinguished from overdefinition of disease. For continuous physiological measures there is a level beyond which some people are deemed ill. Take blood pressure and the continued debate as to how firmly to set the threshold for hyptertension at 140/90 mmHg. Though it’s not even clear whether mild hypertension should always be treated pharmaceutically some would have us now fear anything above120/80 mmHg as prehypertension. These sliding brackets of course define more and more people as ill, though we are talking about relative risk of disease rather than any explicit pathology. But how is that threshold determined and who gets to set it? That question will be left for another issue.
Perhaps the most extreme case study of overdiagnosis is that of screening for thyroid cancer in South Korea. From 1999 the government offered thyroid ultransonography as part of a wider health screen. The detection of thyroid-cancer increased steadily until in 2011 almost 70 per 100,000 Koreans were diagnosed, making it the most commonly diagnosed cancer in that country. But although the apparent incidence of thyroid increase 15 times the pre-screening level of 1993, the numbers dying from the disease changed very little over this period.
The only explanation for this increase was the detection of small papillary tumors few of which would have caused harm in a patient’s lifetime. But most of those diagnosed with thyroid cancer undergo thyroidectomy requiring lifelong thyroid-replacement therapy. Complications like hypoparathyroidism and vocal-cord paralysis are not uncommon so in many cases it would have been better to have never known about the tumor.
The numbers game and PSA
Alexandra Barratt, Professor of Public Health at the University of Sydney tells of the similar story for prostate cancer screening in Australia. “There was a more than doubling of the incidence of prostate cancer since the PSA test came in 1989 and a lot of that would be overdiagnosis,” she says. “There are a lot of guys out there who think their lives have been saved but actually they’ve been overdiagnosed.”
Everyone knows someone who died because a cancer was detected too late for effective treatment and it’s counterintuitive to think that screening for early growths would not help. “We can’t identify individuals who have been overdiagnosed. It’s something that only visible at the population level;” says Professor Barratt, “which means that we have to rely on epidemiological studies to know that it exists. So it’s taken some time to accumulate an evidence base to show that.”
Prostate cancer is the most diagnosed cancer in Australia with an incidence of 25 per cent; effectively 17 250 new cases in 2015. According to modeling data if 1000 men between the ages of 55 and 69 were left unscreened, five would die of prostate cancer and 55 would live with asymptomatic cancer1. However, if annual screening were implemented one or two deaths would be avoided by early intervention, but there would be grave complications for many more men.
Twenty-eight men would suffer side effects following a biopsy that required healthcare or hospitalisation and three times this many would learn after biopsy that their PSA result was a false positive. Thirty-seven men with elevated PSA would be found to have slow-growing cancers not requiring treatment but 25 would choose to have therapy just to be on the safe side.
Seven to 10 men who have treatment would experience more serious complications such as impotence, urinary incontinence or bowel problems, and one in 2000 could have a heart attack due to treatment. Given the burden of therapy, cancer diagnoses are also associated with reduction in work hours and up to 40 per cent of income for some period according to a study published last week Cancer.
Despite the high rate of overdiagnosed cancers it’s very hard to shake the feeling that screening must be helping by getting the clinically significant ones early. This perception is reinforced by the problem of ‘lead-time bias.’ Imagine two men die from prostate cancer though one of them was diagnosed two years earlier by a PSA test. Although it appears that he survived longer from time of diagnosis because of screening, the result is little more than a numerical tautology.
At any point in time there will be a spectrum of cancerous growths at various stages of progression. Professor Barratt explains how, “the slow growing tumors just hang around waiting for you to have a screen and pick them up. The fast-growing ones are likely to go all the way from preclinical to metastatic in the period between the screening intervals; they’re the ones you’re mostly likely to miss.” A third of men over 50 who die of causes other than cancer have been shown in post-mortem examination to have small, indolent prostate cancers and this proportion doubles for the over 60s.
Too much information
The 5-year survival from prostate cancer is now at least 93 per cent in Australia. So with all these numbers floating around, don’t epidemiologists have clever algorithms like ‘numbers-needed-to-treat’ that can absolutely determine whether screening is worthwhile for public health and how often it should be done? “It’s a much more nuanced procedure working out whether making the diagnosis itself is worthwhile [than it is to evaluate therapeutic interventions],” explains Professor Glasziou, who has sat on boards both at MSAC and the PBS. “Getting the bits of evidence to link up is hard.”
It is ironic that improvements in technology have led us further into this trap. The blood tests and the x-rays work almost too well by revealing pathology we don’t need to see. Professor Paul Glasziou gives the example of pulmonary embolism. “Spiral CT detects such small blood clots in your lungs that they are very low risk or probably harmless. One large randomised trial compared Spiral CT to the old technology and found no difference overall in the mortality for the pulmonary embolism from the two groups. But the spiral CT detected more pulmonary emboli and those people are now being anti-coagulated,” says Professor Glasziou.
In the case of cancer screening, “what you want is to pick up the cancers that really matter rather than picking up masses of indolent cancers,” says Professor Alexandra Baratt. “Unfortunately this isn’t going to be solved overnight because we don’t have the biomolecular knowledge that we need to identify those cancers that are going to be progressive and those that won’t. I’m sure we will in time,” she concludes.
Looking a patient in the eye
For all the evidence showing that PSA testing leads to more harm than good, how do you explain this to a patient? And no doctor can tell whether the patient in front of them will be one of the overdiagnosed or one of those that will fall victim to a malignant cancer. With conflicting guidelines the ball is left in the doctor’s court, and it’s an agonising ball to play.
RACGP guidelines recommend that doctors do not discuss PSA test unless the patient specifically asks for it and that they should be able to explain the complicated costs and benefits. But Professor Barratt’s research shows that only about half the community has some understanding of overdiagnosis although none associated this as a risk of screening. Further, Australian survey data published this month indicate that the public prefer GPs to initiate discussions about PSA testing in asymptomatic men over 502.
In practice doctors behave in very different ways with the same information at hand. Kristen Pickles is a researcher at the Centre for Values, Ethics and the Law in Medicine, at the University of Sydney, who has investigated the factors that influence decision-making by GPs on PSA testing.
“Many GPs used personal or professional experiences with the PSA test, both positive and negative, as powerful anchors for their current practice,” states the article in BMJ Open3. Doctors surveyed often reported a sense personal responsibility or guilt at having missed a cancer diagnosis in time to prevent a patient’s death, or the possibility of doing so.
The qualitative study showed that GPs followed one of four patterns of behavior. Some doctors trusted that the RACGP guidelines for best practice were appropriate and would protect them from medicolegal risk. They would explain the pros and cons of PSA testing and accepted that observing occasional adverse outcomes was “just part of being a GP, you have to walk around with this.” Another GP told the researchers, “it happens. I mean, unfortunately, no matter how good a doctor [I am], now and then this is going to happen.”
Other doctors took the approach that certain individuals might not be capable of understanding such complex information and they tailored their discussion appropriately. Or they didn’t discuss the evidence base at all, feeling it was the responsibility of urologists to explain the details. But Kristen Pickles says that “GPs are the gatekeepers to the PSA test and everything that might follow, so they need to understand [the findings]” and be able to explain them to patients.
Many GPs still stick with the ‘better-safe-than-sorry’ mentality of always recommending a PSA test. “GPs who worried about medicolegal risk feared missing diagnosing cancers. They were wary of guidelines and authorities advising against screening,” says Kristen Pickles. Conversely, some adopted the opposite approach to avoid overdiagnosis at all costs. One GP reported that although PSA testing saves some lives, “the problem is, it just doesn’t happen often enough to balance out…all the damage that we do.”
These ‘extreme’ stances amongst doctors may well be the norm. A population survey published by Barratt and colleagues showed that only 18 percent of men reporting participation in prostate cancer screening said they’d been told about the risk of overdiagnosis at some point4. For women undergoing breast screening the figure was even lower at 10 per cent.
Elizabeth Dawson is a British woman who started a campaign against routine screening after having had a mastectomy within months of a positive breast screen. “She described her experience of being invited to be screened as though it were of no more significance as going for a dental checkup,” says Professor Barratt. “She felt very let down by the screening service that she hadn’t been given adequate information about it.
“She contrasted this to when she was pregnant and the system put a lot of time and effort counseling her about whether to have a prenatal test for Down Syndrome and to understand the consequences of what having that test would be,” describes Professor Barratt.
It’s a huge responsibility to land GPs with, and in order to give them the confidence to open such conversations with their patients, Kirsten Pickles says, “GPs need active support to implement more conservative management strategies in line with the evidence.” Inevitably some patients will have limited health literacy and may “trust in their GP to make an appropriate or ‘best’ decision and explaining it to them, but it does raise many questions about who should be [medicolegally] responsible for taking on the burden of this complex situation.”
Professor Glasziou has been gauging public reaction in ‘community juries’ in which he presented the risks of PSA screening while a urologist would present the benefits. “Their eventual decision was that they were unenthusiastic about being screened and that there should be no national campaign for screening but it takes a while,” he says. “It’s not a superficial thing of simply presenting some numbers to them. It actually takes a lot of back and forth to digest this.”
This laborious educational process is not feasible in the brief window of a single GP consult. Doctors can instead guide their patients towards resources such as the RACGP-developed patient information sheet on prostate cancer. This give patients time to digest the information and to take on some responsibility for this critical decision. It is a deeply personal choice since risk will differ from case to case based on family history and ethnicity. For example, black men have twice the lifetime risk of prostate cancer and associated mortality as Caucasians according to a recent British study5.
“Optimal care falls somewhere in between GPs making case-by-case decisions and GPs avoiding overdiagnosis. This is shared decision making,” says Kristen Pickes. “Involving patients decision-making while practicing according to the evidence base is less of a lottery than treating each patient the same.”
Breast screening back under the spotlight
Cancers diagnosed by biennial mammogram screening were likely to be bigger and more advanced than those detected by annual screening according to a study published last week in JAMA Oncology1.
The eight year prospective study of 15 440 women found that biennial vs annual screens revealed a higher proportion tumors is stage IIB or above (26 per cent vs. 20 per cent), tumors larger than 15 millimeters (65 vs. 55 per cent) and nodepositive disease. For postmenopausal the screening frequency made no difference on these measures.
Although the authors did not determine whether the differences in screening rates would affect breast cancer mortality the publication coincides with changes in the American Cancer Society’s guidelines that recommend women annual screening mammography from the age of 45. Women 55 years and older can consider screening biennially and routine clinical breast examination is no longer recommended.
The current research reignites the on-again off-again debate over the relative efficacy of breast screening. Only last July a systematic review had cast doubt on the effectiveness of breast screening at reducing mortality. While the original Swedish trials of the 1960s and 1970 had found that screening could prevent a quarter of breast cancer deaths a review of these data showed that the figure was no more than 10 per cent2. The authors of that study pointed out that the reduction in deaths since routine mammography was introduces was largely related to better treatment rather than screening.
In similar metanalyses of screening trial data, Professor Alexandra Barratt and colleagues found that “around 15 to 20 per cent of the breast cancers diagnosed among the women in the screened group during that time are overdiagnosed.” But contrary to the widespread understanding of PSA testing “it’s very counterintuitive to think that there might be breast cancer overdiagnosis because we were not trained like that,” she says.
Many cancers detected are Ductal Carcinomas In Situ (DCIS) that never invade beyond the mammary ducts. Others are progressive but so slowly growing that they never cause symptoms or become metastatic. “They didn’t need to be found, and didn’t need to be treated” says Professor Barratt. Of course treatments involving surgery, radiation, chemotherapy and hormonal supplementation are fraught with their own complications.
Until better biomarkers are available to identify the aggressive breast cancers Professor Barratt first suggests a risk-based screening frequency to minimise the frequency of overdiagnosis. “So if you have a low family history and not dense breasts, you may choose to screen less frequently, perhaps every three years and then five yearly.” Women with higher risk factors could continue screening every two years as is currently recommended, she says.
Professor Alexandra Barratt also thinks that a less knee-jerk approach could be adopted once tumors are found. In the British LORIS trial, “they are randomising women who are diagnosed with low-risk DCIS to active surveillance and no active treatment or to immediate treatment to see whether there is a place for not hitting the ‘trigger’ button.”
Apart from the side-effects of therapy the mere use of the term ‘cancer’ can of its own accord cause great anxiety among the diagnosed. A study published this month in Breast reported that “communicating a diagnosis of DCIS using terminology that does not include the cancer term was preferred by many women and may enable discussions about more conservative management options.”
Professor Paul Glasziou explains how the mere stigma of a cancer diagnosis can cause anxiety among the ‘worried well’. “We really need to start developing 21st century terminology because cancer is no longer what we thought it was. Now we’re detecting things that are so early and so indolent and so benign that using the term cancer for the whole lots leads to a huge misinterpretation.”
The age-standardised incidence of breast cancer in this country is 115 per 100,000 women. Between 2013 and 2014, over 1.4 million women aged 50-69 had a mammogram through BreastScreen Australia which covers about 56% of the target population. While women can have a mammography with Breast Screen Australia without a referral, “women value what their GPs say,” says Professor Alexandra Barratt. As for prostate cancer screening in men, general practitioners have the difficult task of keeping up with ever-changing numbers and making recommendations to their patients.