“Don’t believe in slogans about cancer screening,” warns Peter Gøtzsche, Professor of Clinical Research and Analysis and director of the Nordic Cochrane Centre. “They are all wrong, misleading or doubtful.”
For some years Peter Gøtzsche kept a folder labelled Dishonesty in breast cancer screening on top of his filing cabinet, storing articles and letters to the editor that contained statements that he knew were dishonest. Eventually he gave up on the idea of writing a paper about this collection, as the number of examples quickly exceeded what could be contained in a single article. He wrote a book instead.
In chapter two of his ground-breaking book, Peter Gøtzsche describes very clearly and succinctly what the important issues in cancer screening are, and outlines the major scientific disputes surrounding mammography screening.1
What it means “to have cancer”
While cancer is a malignant disease that consists of abnormal cells that spread uncontrollably throughout the body, it isn’t always fatal. Yet the public image of cancer is that if left untreated it will kill you. However, recent data suggests that a substantial fraction of detected cancers and precursors of cancer will not prove fatal and will regress. They include cervical cancer, colon cancer, renal cell cancer, melanoma, breast cancer, and neuroblastoma.
Cancer is so common that it is likely that all middle-aged people have cancer somewhere in their body. It is a biological fact of life that we cannot avoid getting cancer as we get older because the telomeres in our DNA, which are like a repair kit that protect us from developing cancer, “runs out of supplies” and cancer cells are then able to develop.
We use the term “having cancer” to mean being ill with cancer, but it means something very different in the context of a screening programme. Because it is possible to detect cancer in virtually everybody over a certain age if we look hard enough, “having cancer” may mean simply having cell changes that will not result in any symptoms or harm us for the rest of our lives. We will die with the cancer but our death is not caused by the cancer we carry. Prostate cancer is one common example of this.
Overdiagnosis and overtreatment
The purpose of screening for a specific cancer is to reduce the mortality rate from that type of cancer. As cancer comprises a group of very different diseases, and many cancers grow very slowly or not at all, one of the important things to know about cancer screening is that it inevitably leads to harm in the form of overdiagnosis and subsequent over-treatment. Overdiagnosis is the detection of a cancer that would not otherwise have been identified clinically during the person’s remaining lifetime.
The problem is that at the moment science cannot distinguish between the harmless cancers that have no symptoms and will not cause any harm and those that are dangerous. So the response to this dilemma is to treat them all. Overdiagnosed breast cancers that would not have caused any problem are treated by surgery, many are treated with radiotherapy, and some with chemotherapy.
As cancer screening always causes harm, it is essential to ensure that the screening being performed does actually reduce mortality from the cancer and whether the reduction is large enough to justify the various harms inflicted on the healthy population that is being screened.
Another important issue in cancer screening is the tendency to over-estimate how accurate the diagnostic tests are. While erroneous diagnoses cause much less harm than overdiagnosis, they are still one of the significant causes of harm that screening results in. Peter Gøtzsche notes that this problem has received little attention in the scientific literature and virtually none in the debates about breast cancer screening.
For example, there are two uncertainties involved in deciding whether a woman has breast cancer. They are reading the mammogram and interpreting the biopsy. These result in some women being diagnosed with breast cancer at screening when they do not have breast cancer, yet there are no publications estimating how often women are wrongly diagnosed with breast cancer.
Carcinoma in situ
Carcinoma in situ is another problem. Peter Gøtzsche calls it a misnomer. Cancer is an invasive disease whereas carcinoma in situ is Latin and means cancer at the site. It is not invasive: the cell changes remain where they originated.
Issues in cancer epidemiology
There are some essential issues that need to be understood when talking about cancer screening.
The incidence of a disease is the number of new cases occurring in a certain time period, often the calendar year.
The prevalence of a disease is the proportion of a population that has the disease at a given point in time. The prevalence includes all cases, including those that have existed for many years. Like incidence, it is usually expressed per 100,000 people and for a certain age group.
Cancer screening primarily detects slow-growing cancers. This is because the longer the cancer has existed, the greater the chance that it will be picked up at a screening session. In contrast, a cancer that grows very quickly is much more likely to be detected clinically, between two screening sessions. These are referred to as interval cancers and they are therefore more dangerous than cancers detected by screening.
Screening increases the prevalence of cancers with an excellent prognosis due to overdiagnosis, and detects most of the slow-growing cancers. This is why cancers diagnosed in a region with screening will have a more favourable prognosis, on average, than cancers in a region without screening. The screened region has its numbers of breast cancer cases boosted by the many harmless cancers that weren’t detected in the other region because the cancers have no symptoms. This problem is called “length bias.” Overdiagnosis is therefore a special form of length bias.
Another common problem is the bias that occurs in comparing regions with and without screening if researchers use the number of years that patients have survived from their date of diagnosis as the outcome. This is known as “lead-time bias.” The problem can be avoided by comparing the mortality rate of each region or country rather than the years of survival.
Peter Gøtzsche states that cancer charities and cancer researchers publish misleading survival analyses so often that it appears like a deliberate strategy to deceive the public into believing that important progress is being made in the fight against cancer.
Breast cancer screening has now become one of the greatest controversies in healthcare today. The general public remains largely unaware of this as media coverage on breast cancer screening chooses to ignore the complex issues as well as the vested interests surrounding breast cancer screening and treatment.
Even more reprehensible is the lack of accurate information given to women by government agencies such as Ministry of Health and the National Screening Unit. It is long past time for the limitations and significant harms of breast cancer screening to be clearly acknowledged so that women are able to make adequately informed decisions before they agree to being screened.
Peter Gøtzshe. Mammography Screening: Truth, Lies and Controversy. 2012. Published by Radcliffe Publishing Ltd.