In August 2018 the British Medical Journal published a lengthy review article1 dealing with the issue of over-diagnosis. This is defined as the diagnosis of a condition (often subsequently treated) that would otherwise not cause symptoms or harm to a patient during his or her lifetime.
Over-diagnosis arises in a number of ways, including:
- The broadening of disease definitions via lowering of diagnostic thresholds and the recognition of risk factors as pre-diseases
- Greater use of technology and technological advances that improve testing sensitivity
- More widespread screening programmes
- Physicians’ over-estimation of the benefits of treatment in mild or low-risk disease
- Changing culture in healthcare, in which patients and also physicians are troubled when problems are not labelled with a diagnosis and some sort of therapeutic action taken.
Examples are not hard to find. Diagnostic thresholds for diabetes in terms of HbA1c or fasting plasma glucose have been on a downward trend for years. A high proportion of adults aged 50 and over are now on a statin targeted at fairly modest elevations of serum cholesterol. (Many are on anti-hypertensive treatment too, although the case for treating more minor elevations of blood pressure might be a little more compelling.)
But it is in the increased use of sensitive screening for cancer that the problem of over-diagnosis is perhaps most serious. The article’s authors claim that population cancer screening programmes in developed countries can result in significant proportions of over-diagnosed cases among total diagnoses:
- Breast cancer via mammography – up to 10% or more
- Thyroid cancer via increased frequency of imaging – up to 49% in men and up to 83% in women
- Prostate cancer via prostate-specific antigen (PSA) screening – up to 80% or more according to age and tumour characteristics.
For prostate cancer they give an example of 1,000 men aged 45 to 80. If none were screened via PSA, 68 would be diagnosed with prostate cancer; of these, seven will die and 61 will survive after successful treatment. With PSA screening, 88 will be diagnosed; of these, seven will die from prostate cancer and 81 will survive after treatment. So the number of deaths remains the same but 20 of the 81 men will have had treatment for a cancer that would not have caused any harm. They will have been over-diagnosed and undergone needless treatment (with quite possibly troubling after-effects).
It may be that ignorance is, indeed, bliss.
Halting or even slowing the trend of over-diagnosis and the drain on healthcare funds is a big challenge. The notion of decelerating technological change is unrealistic, and there are high cultural barriers to surmount. There are moral ones too: for example, how much is it worth spending on screening to save one life? And having created screening programmes, how easy is it for governments subsequently to restrict their availability?
How does over-diagnosis affect insurers?
While an individual may take some comfort from a diagnosis and the receipt of treatment, he or she now has something to disclose to an insurer. That disclosure may result in a lengthier underwriting process while evidence is obtained and an adverse decision – in the case of a cancer over-diagnosis, even a postponement.
And if special terms are quoted, how fair is that decision? By their very nature over-diagnosed conditions have no or merely a negligible effect on mortality and morbidity. Does a case of symptomless diabetes just over the HbA1c threshold merit the rating for the type 2 condition straight out of the manual? It is likely that in some instances rating recommendations are for cases that have an average severity higher than those seen by underwriters today. This is another reason for those setting ratings to keep up to date in their evidence-based research.
Also, over-diagnosis inevitably shrinks the pool of standard-rates lives: more people with something to disclose, more rated decisions. If improvements in insured-life mortality are seen, how much is real and how much is due to re-classifying what would have been ordinary-rates cases and transferring them to the substandard group?
And insurers need to be satisfied that their routine evidence requirements are reasonable not only in the context of cost-effectiveness but also of anti-selective threat and impact on the applicant. They should also consider whether informed consent is required; for example an abnormal PSA result may have major consequences for quality of life, even if there little or none for mortality and morbidity.
Over-diagnosis is just one medical issue that affects insurance and the way insurers deal with their customers. It underlines the need to be aware of the changing world around us and to adapt accordingly.
- Overdiagnosis in primary care: framing the problem and finding solutions. Kales MS, Korenstein D. BMJ 2018;362:k2820