The measurement and interpretation of prostate specific antigen (PSA) levels in otherwise asymptomatic men has caused a great deal debate in the medical world. Worries about the harms of overdiagnosis and overtreatment of a cancer that is never going to cause significant clinical problems or impact mortality has led to reconsideration of the value of PSA screening, particularly at a population level.

Drives to reduce harm from management approaches such as prostatectomy or radiotherapy have also had an impact. When the interventions may be worse than the outcome of the illness for many there is little argument for screening.

As insurers it is important not only to evaluate carefully what the academic papers report on a subject, but also to understand what is likely to happen on the ground with their clients. The real-life experiences of the insured population in their uptake of screening and subsequent investigations can be understood only in the context of the clinical and social structures in which they live their day-to-day lives.

The UK National Health Service (NHS) has raised a number of concerns about the rise of commercially available screening,¹ particularly when the infrastructure (especially quality assurance) to manage the test results is not in place. Insurance medical examinations may not always be beyond reproach in this case.

This piece, taking lessons from recent consensus discussions,² hopes to give some greater understanding of the pros, cons, limitations and benefits of screening in general, and PSA in particular, while looking for lessons that the insurer can take away from the debate.

The NHS in England describes screening as “a way of finding out if people have a higher chance of having a health problem, so that early treatment can be offered, or information given to help them make informed decisions”. Straightforward enough perhaps, but it begs several questions and has implications for its use in underwriting assessments.³

Screening should be offered only to those who are likely to benefit from such screening. From a clinical perspective early recognition has to offer benefits such as more effective treatments and reducing the chances of complications of the disease (at worst early death, e.g. cervical screening). Although much has been made of potential screening tests for Alzheimer’s disease the lack of an effective treatment suggests that it has little practical advantage beyond the important identification of research subjects.⁴

It should be recognised that, with very few exceptions, screening tests do not make a diagnosis but initiate a chain of diagnostic procedures to test the screen against gold-standard clinical practice. Screening at the underwriting stage potentially offers little or no benefit to the person screened.

The properties of an effective screening programme are set out below.

The Properties of an Effective Screening Programme⁵

• The condition must be an important health problem judged by its severity and frequency with well-understood natural history, incidence, and prevalence.
• There is robust evidence of association between the disease marker and serious or treatable disease.
• The test is simple safe, precise and validated with the distribution of test values known in the population.
• A suitable cut-off level is defined and agreed.
• The time from sample collection to delivery of results should be acceptable to the target population.
• There is an agreed pathway of further investigations for those found to be positive.
• Effective interventions are available with evidence that identification in the pre-symptomatic phase leads to better outcomes than usual care based on interventions with a sound evidence base.
• Benefit gained by individuals should outweigh any harms from overdiagnosis, over treatment, false positives, false reassurance uncertainty over findings and complications.

Potential Difficulties with Screening

Problems arise when screening tests are not clearly related to a gold-standard diagnosis or the cases and control population in which the screen was trialled are not representative of general populations (and by implication not representative of the population where the test will be used).

This raises the question of the generalisability of the test and its interpretation in other populations. Slowly progressive diseases can be over-represented in screened subjects and, most pertinent to PSA, subclinical disease that may never become a problem but is detected by the screening can make the screening tool too sharp an instrument for the task in hand.⁶

The impact of screening for prostate cancer with PSA has been investigated extensively over the years. The most impressive results showed a 20% relative risk reduction in prostate cancer-specific mortality (PCSM)⁷ as a result of screening. By way of contrast, others studies have been less impressive, showing a 0.09% reduction after 15 years following a single test or 3% reduction in localised disease with no statistically significant difference between various treatment arms.^8,9

These pivotal studies raised important questions. First, that there was an over-representation of white men in the study groups. This was unhelpful as it may be the case that screening is more effective for black men, whose lifetime risk of developing prostate cancer is double that of white men.¹⁰

Second are the problems associated with overdiagnosis and overtreatment. Concerns that mild and sub-clinical cancers identified by screening were managed with interventions that potentially caused significant harm (with no improvement in overall outcome) led to changes in both diagnostic and management pathways. These problems may be both short and long‑term.¹¹

The use of magnetic resonance imaging and PSA density (calculated dividing the PSA by the volume of the prostate gland) has been shown to reduce the need for more invasive biopsy by 50%.¹² MRI-guided biopsy has led to significant improvements in recognising low-grade cancers that do not need intensive treatment. Active surveillance in conjunction with these techniques allows some people to delay or avoid all together radical treatments without compromising the outcome.¹³

The time needed to understand the benefits of changes in screening or management strategies is, by medical studies standards, very long – perhaps 10 to 15 years. The logistics of follow‑up studies are challenging but without this data agencies can find it difficult to recommend changes in screening protocols. By comparison to the term of many insurance policies and the evidence that would be most useful to inform evidenced-based rating, this is not a particularly long time.

The recent UK clinical consensus document concerning screening asymptomatic men with PSA fell short of recommending that all men over the age of 50 should have a PSA screen. Greater stress was given to making the test available to those who had been properly counselled about the implications of the test, with a particularly forceful recommendation to proactively raise the issue (not currently recommended in UK guidelines)¹⁴ with those at higher risk, including black men, and those with a family history or the BRAC2 gene.

This raised the need for proper training across a number of agencies but recognised that there were significant challenges in making the aspiration a reality within the healthcare system, particularly in primary care.

A recent expert panel in the U.S. recommended annual PSA screening for black men over the age of 40.¹⁵ The earlier 2018 U.S. guidelines recommended “shared decision-making” for men aged 55 to 70 with a life expectancy of more than 10 years. PSA screening has declined in younger white and black men (aged 40 to 54 years) over the years but the decline was greatest among younger black men, suggesting the message has not been heard.¹⁶

Attention has been drawn to the myriad guidelines worldwide and the differences of their recommendations. Some of the inconsistency has been ascribed to the professional alliance of the writers (urologists vs public health specialists) or the nature of the health insurance system (public vs private).¹⁷

The rationale behind sending all men aged 50 to 69 with a PSA >3.0ng/ml for further investigation was questioned, but there was no agreement on an alternative threshold.¹⁸ It is recognised that black men without prostate cancer have higher PSA levels than white or Hispanic men but the diagnostic accuracy of PSA in men of different ethnicities is not known. This lack of information concerning the impacts of age and ethnicity on PSA levels leads to further ambiguity. Current guidance for interpretation of PSA results does not account for ethnicity.¹⁹

The panel was unable to identify any validated risk stratification tools that could help guide practitioners (family history and BRAC2 status were rarely available) or give guidance on the frequency of testing. Currently frequency of re‑testing relies on the clinical judgement of practitioners. Throughout the paper the authors returned to these as a central problem that needed to be addressed by the academic community.

Work continues into understanding the combination of factors that should lead a person down the pathways to further diagnostic work up, intervention and improving the risk/benefit ratio of engaging with any screening programme.²⁰ This work included the identification of more effective biomarkers of cancers requiring intervention.²¹ Risk-adapted PSA screening, tailored to the person’s individual circumstances, within the clinical arena will no doubt usefully continue.

Insurers continue to ask for screening tests to be performed as part of the underwriting process to reassure themselves that the applicant does not harbour any covert disease that has the potential influence their risk profile. These tests are frequently not clearly indicated from a medical perspective and can be difficult to interpret without a full clinical context. When these results are within the normal range they present no particular problem and pass unremarked. When they are abnormal, then challenges arise.

Applicants may not be adequately counselled about the potential implications of the results obtained. Their doctors may not be aware that these tests are being carried out but become saddled with the burden (emotional and financial) of explaining and managing the fallout, even when the results have no clinical implications. The insurers have a duty to ensure that any tests and the processes of obtaining them are fit for purpose.

The example of PSA leads to the conclusion that insurers need to be mindful that their screening requirements adequately meet the criteria referred to above, and are undertaken within a proper ethical framework, if they are not to provoke significant anxiety for the applicant and cost to associated healthcare systems.