How a population-based cohort of men estimate lifetime risk of prostate cancer in a survey before entering a prostate cancer screening trial in Sweden?

Discussion

In this population-based study with more than 13 000 participants, we found that the median of men’s perceived lifetime risk of prostate cancer was approximately 30% which was significantly higher than the average risk in a screening population (20%). Only one in five men estimated their risk between 15% and 25%, and three out of five estimated their risk to be higher than 25%. Hence, on a population level, there is an overestimation, even though for instance men with a family history of prostate cancer are correct in estimating their risk higher than men at average risk. Several factors were significantly associated with perceived risk, including family history of prostate cancer, LUTS and ED which were associated with >5 percentage points higher risk estimation.

To the best of our knowledge, this is the largest survey investigating how men estimate their lifetime risk of prostate cancer. Other strengths include the high response rate (73%) and the fact that men completed the survey at home before the screening procedures. The main limitation is the risk of selection bias since the study population consisted of men who had chosen to participate in a prostate cancer screening trial. These men could potentially perceive their risk differently than the general public since risk perception affects health-seeking behaviour.16–18 Interestingly, non-attenders’ risk estimation was almost the same as attenders’. The median and upper quartile are identical, while more men among attenders predicted their risk low than among non-attenders as can be seen in figure 3. In the linear regression, the statistically significant factors for non-attenders were also significant for attenders. However, more factors were non-significant for non-attenders. This is partly due to the smaller sample size leading to lower power. For instance, the point estimates for comorbidity were higher for non-attenders than for attenders, although not statistically significant. Altogether, the results show that risk estimation among attenders and non-attenders are similar, indicating that these results can be generalised to men of the same age and background.

Most previous studies investigating perceived risk of prostate cancer have asked men to estimate their risk in categories such as low, average, high10 19 20 or as a relative risk compared with the average man10 21–24 rather than a numerical value. Two exceptions were found. The first is a small Swedish study where unaffected men in families with hereditary prostate cancer estimated their own risk to be 50% of receiving a prostate cancer diagnosis and the risk for the general male population to be 33%,12 which is much in line with the estimates reported herein. The second is a telephone interview-based, Australian study in which men estimated their own lifetime risk of prostate cancer to be 25% and the risk for an average man to be 32%.25

Risk assessment is for many people a difficult concept. A limitation of this study is therefore that the risk assessment was done only by a one-item numerical rating scale from 0% to 100%. In previous studies in breast cancer, this type of measurement often resulted in overestimating the risk, because the anchors were sometimes misleading, and some women perceived their risk of developing breast cancer as 50%, not understanding that this means a one in two chance.26 However, we acknowledge that there is no gold standard measurement for perceived risk and the one-item numerical rating scale is a validated instrument with acceptable psychometric properties that outperforms other items and formats and is frequently used.27 28

There were other questions regarding how well men understood the concept of lifetime risk. Many men probably interpret lifetime risk of prostate cancer as the lifetime risk of developing symptomatic prostate cancer, not being aware that participating in PSA screening also increases the risk of overdiagnosis of asymptomatic cancer. This speculation is supported by the fact that men who choose not to have a PSA-test estimated their risk almost identical to men who were PSA-tested. The lifetime risk of symptomatic prostate cancer before the PSA-era has been estimated to approximately 11%–13% which means that if men interpreted the question as lifetime risk of symptomatic prostate cancer, the overestimation was even larger (figure 3).29–31 We based the anticipated risk of prostate cancer on the observed 24 year cumulative incidence of prostate cancer in the screening group of the G1 trial in order to have a risk estimate from a screening setting. Even though this is not a lifetime risk since not all men in the G1 trial have died, we believe that it is a fair estimate since the incidence curve has flattened out due to deaths from competing events (unpublished data).15 Figures similar to our estimate have also been published by others.29 32

We can only speculate why men in our study overestimated their risk of prostate cancer. Public campaigns aiming at raising awareness about prostate cancer and the invitation letter mentioning ‘prostate cancer is the most common cancer among Swedish men’ may influence men’s perception. Family history of prostate cancer is a well-known risk factor for prostate cancer and in some studies, it has been associated with a two- to threefold higher risk for prostate cancer.33–35 The higher estimated risk in men reporting a family history of prostate cancer could therefore be an adequate risk estimation for these men. Men with symptoms (LUTS and/or ED) also estimated their risk higher. This finding is not surprising since men often associate urinary symptoms with prostate cancer and is often included in patient information materials.36 For example, on the official information website for the Swedish healthcare system, frequency, a weak stream and urinary hesitance are listed as typical symptoms of prostate cancer.37 Most men with LUTS however, do not suffer from prostate cancer but rather have benign prostate enlargement and the association between LUTS and prostate cancer is weak.38–40

The inconsistency between the perceived and actual risk of developing prostate cancer is an important finding for several reasons. First, since higher perceived risk is linked to higher worry for cancer,41 men’s overestimation could mean that they worry about prostate cancer unnecessarily and improved information could potentially reduce this anxiety. Second, if men do not have an accurate risk estimate, they cannot make a truly informed decision as to whether to be screened for prostate cancer or not. Third, given our results that men associate LUTS with prostate cancer risk, this is a group that particularly could benefit from improved information. Many men may worry about prostate cancer since LUTS are very common.42 These men may decide to have a PSA-test based on the assumption that they are at risk for prostate cancer, while those without urinary dysfunction may abstain from testing themselves in a false sense of security, believing that they are not at risk for prostate cancer, potentially leading to a delayed diagnosis.

Our findings show that men overestimate their risk of prostate cancer, and thereby potentially also overestimate the benefit of screening which is in line with other reports showing that both patients and healthcare workers tend to overestimate the benefits from medical interventions.43 44 Thus, there is a need to improve information concerning prostate cancer risk, risk of overdiagnosis and outcomes after medical interventions helping men to get a realistic risk perception. Prediction models may help to obtain personalised risk estimates. Such improved information could be communicated using icon arrays and numerical risk estimates, so that men can make a well-informed decision on whether to be screened or not for prostate cancer.45 Urinary symptoms is a common reason why men ask for PSA-testing but the link between urinary symptoms and prostate cancer should be removed36 40 and the public messaging should instead emphasise that prostate cancer commonly is silent or asymptomatic, particularly in the curable stages of the disease. Men with urinary symptoms should have the same information regarding benefits and harms of screening for prostate cancer.