Associations between socioeconomic factors and PSA testing in a population-based organised testing programme and routine healthcare: a register-based study of 50-year-old men

The OPT programme

The OPT infrastructure and diagnostic pathway have been previously reported in detail.16 In brief, the Region Västra Götaland (RVG) OPT project since 2020 actively invites all 50-year-old men in the region. The public was informed in local and regional newspapers, on social media, at bus stops, and through a broadcasted live information and question session. A nationally standardised information text about the benefits and harms of PSA-based prostate cancer testing is included in a three-page invitation letter, described in detail elsewhere.17 This information was at the time of this study translated into English, Dari, Persian, Arabic and Somali and was made available as a video with sound and sight interpretation. Geographically selected testing facilities at public hospitals and primary care centres, were available for the men to obtain a PSA test. The number of facilities increased from 27 to 36 over the study period. Information about the testing facilities was included in the invitation letter and available on a website. The distance to the nearest facility was usually less than 5 and never more than 25 km. The PSA blood sample was specifically labelled to link it with OPT participation. The men could use their invitation letter to obtain a PSA test within 30 days. They could extend this time if they contacted the OPT office. The OPT PSA test was free of charge, but men who had an MRI scan, with or without a subsequent urology appointment, were charged 300 Swedish kronor (approximately €25).

OPT: population and data sources

This study included all men invited in 2020 and 2021 to the RVG OPT programme. Data about participation (obtaining an OPT PSA test) were obtained from the regional OPT database at Regional Cancer Centre West. After approval from the Swedish Ethical Review Authority, socioeconomic data for all invited men were collected from Statistics Sweden’s Longitudinal integrated database for health insurance and labour market studies (LISA, by its Swedish acronym).

Unorganised PSA testing: population and data sources

Data for the unorganised, pre-OPT PSA testing were collected from the Western Sweden Study of Opportunistic Prostate Cancer Screening (WSOP) database. WSOP includes all men aged over 18 years who were a resident in RVG at any time between 1 January 1995 and 31 December 2014.18 19 WSOP includes the date and result of all PSA tests and prostate biopsies obtained in RVG, the date of any prostate cancer diagnosis and linked data from several national registries including the LISA database and the Swedish Cancer Register. This study included men who were 50–52 years old in 2013–2014 and did not have any previous prostate cancer diagnosis. The 51-year-old men and 52-year-old men were included to increase the statistical power. For men with more than one PSA test during the study period, only the first PSA test was included in the analysis. The assay used for PSA analysis was the same in the unorganised setting and in OPT. During the studied years, the Swedish Board of Health and Welfare recommended that PSA testing of men without symptoms should be done only after informed decision-making, including the presentation of the national standardised information about potential benefits and harms of PSA testing (this recommendation still applies).

Socioeconomic categorisation

The analysed socioeconomic factors were education, individual income, cohabitational status and country of birth. Educational level was defined as the highest completed education and categorised as primary school (<11 years), secondary school (11–12 years) or tertiary education (high school/university, >12 years). Income was categorised as low (<50% of median income, the standard definition used by Statistics Sweden) or not low (≥50% of median income). For the OPT population, both adjusted gross income and disposable income were analysed. The WSOP database does not include data on disposable income, so for the pre-OPT population, only adjusted gross income was analysed. Cohabitational status was categorised as single-person or non-single-person household, and country of birth as Nordic (Sweden, Norway, Denmark, Finland or Iceland) or non-Nordic.

Statistical analysis

The outcome measure in the OPT population was participation, defined as having a PSA result within the OPT programme; in the unorganised, pre-OPT population it was having a PSA result obtained in RVG between 1 January 2013 and 31 December 2014 without a previous prostate cancer diagnosis. Univariable and multivariable incidence rate ratios (IRRs) with 95% CIs for these outcomes were calculated with Poisson regression, using the most common category in the source population as a reference. Interactions between education and family type versus the other socioeconomic factors were described with cross-tabulation. The unorganised, pre-OPT population was analysed to establish the testing patterns prior to the OPT. No statistical comparison between the OPT and pre-OPT populations was done because the reasons for accepting an offer for OPT and for having a PSA test in the pre-OPT period (eg, obtaining medical care for lower urinary tract symptoms) were quite different (ie, a null hypothesis was deemed inappropriate). A post hoc descriptive analysis of the PSA testing results was conducted to explore the differences between testing in the OPT versus the pre-OPT setting.

Patient and public involvement

1. At what stage in the research process were patients/public first involved in the research and how?

   Patient organisations and public representatives were engaged in the writing of the OPT information letter and the development of informational material.

2. How were the research question(s) and outcome measures developed and informed by their priorities, experience and preferences?

   Socioeconomic equity is highlighted as an important aspect in the recommendations from both the European Commission and the Swedish Board of Health and Welfare. Both these organisations represent the public.

3. How were patients/public involved in the design of this study?

   No patient or public representative was involved.

4. How were they involved in the recruitment and conduct of the study?

   Not applicable.

5. Were they asked to assess the burden of the intervention and time required to participate in the research?

   Not applicable.

Participation in the organised testing programme

In 2020 and 2021, 21 174 men aged 50 years were offered to participate in the OPT programme, of whom 7871 (37.2%) participated by having a PSA test within the programme. The PSA reading was over the cut-off of 3 ng/mL for 2.2% of the men (95% CI 1.9% to 2.5%). The proportion of men with PSA≥3 ng/mL per socioeconomic category is shown in online supplemental table 1.

OPT participation rates ranged from 20.5% in men with low disposable income to 43.5% in men with tertiary education (table 1). Low education, low income, non-Nordic country of birth and living in a single-person household were all negatively associated with the participation rate (table 1). Income was the factor with the strongest association on both univariable and multivariable analysis (multivariable IRR 0.63, 95% CI 0.58 to 0.68). Cross-tabulation showed that single men with low income (16.9%) were the least likely and non-single men with tertiary education (46.3%) the most likely to participate in OPT (online supplemental tables 2 and 3).

Unorganised PSA testing prior to the organised testing programme

A total of 32 989 men who in 2013–2014 were aged 50–52 years and had no prostate cancer diagnosis were identified in the WSOP database. Of these men, who were not invited to PSA testing but should be tested on clinical suspicion of prostate cancer and could be tested on their own initiative in the absence of symptoms, 4670 (14.2%) had one or more PSA tests taken in these years. The first PSA reading was over the cut-off 3 ng/mL for 5.7% of the men (95% CI 5.0% to 6.4%). The proportion of men with PSA≥3 ng/mL per socioeconomic category is shown in online supplemental table 1.

The proportion of men having a PSA test ranged from 11.8% in men with low income to 15.1% in men not living in a single-person household (table 2). In this pre-OPT setting, only income and cohabitational status were significantly associated with PSA testing (table 2). Like in the OPT setting, income was the factor with the strongest association on both univariable and multivariable analysis (multivariable IRR 0.83, 95% CI 0.78 to 0.90). Cross-tabulation showed that single men with low income (11.2%) were the least likely and non-single men with secondary education (15.4%) the most likely to have a PSA test (online supplemental tables 4 and 5).

Socioeconomic factors and participation in prostate cancer screening trials

The overall participation rate among 50-year-old men after a first invitation to the Swedish regional OPT programme was 37%. This is in between the reported rates from randomised screening trials: 12% in the first round of the German PROBASE trial, which invited 45-year-old men,20 and 59% in the first round of the Swedish Gothenburg-1 trial, which invited men aged 50–69 years.21 Participation in prostate cancer screening trials is often defined as participating at least once during a longer period, which is not comparable to our results from a first invitation to the youngest individuals in the target group. In a pilot study of OPT in the southernmost part of Sweden, younger men were less likely to participate: 38% of 50-year-old men participated compared with 44% of 56-year-old men and 45% of 62-year-old men.22 Several other factors may influence participation such as being invited to a research trial as opposed to standard healthcare like the OPT programme, the information in the invitation and the use of reminders to non-participants. If a national screening programme was implemented, participation would likely be greater because of the healthcare authorities’ recommendation of it, and because of subsequent greater general awareness of prostate cancer screening.

The association between socioeconomic factors and participation in prostate cancer screening trials has been reported only from a Finnish trial, in which low education and low income were associated with lower rates of overall participation by 13.7 and 13.6 percentage points, respectively.10 These differences are somewhat smaller than those that we observed in the Swedish OPT programme. Possible explanations include the different age groups and settings (screening trial vs informed decision in standard healthcare).

Socioeconomic factors and participation in established cancer screening programmes

Although the Swedish regional OPT programmes, including the one in RVG that formed the basis for the present study, are in many ways similar to a formal screening programme, there are some important differences. A formal screening programme is usually recommended by a healthcare authority; in contrast, the invitation letter to the RVG OPT programme makes it clear that the Swedish healthcare authorities do not recommend a national screening programme for prostate cancer, as the advantages do not clearly outweigh the harms, but that they advise that men should be informed and make a personal decision whether to obtain testing.

Established screening programmes for breast, cervix and colon cancer have reported socioeconomic gradients for participation.13 23–30 In Swedish breast cancer screening, the combination of low income and living alone was strongly associated with low participation,31 which we also observed in our study of prostate cancer testing. Other factors negatively associated with participation in that study include not owning one’s home and having social assistance or benefits as the main source of income.31 None of these factors were included in our analysis. In most studies, low income, low education, living alone and being an immigrant are associated with lower participation. A meta-analysis of 66 studies reported ORs of attendance to breast cancer screening of around 2 for high versus low income, married/cohabiting versus unmarried/non-cohabiting and immigrant versus non-immigrant, but only 1.1 for high versus low education.24 A global literature review of participation in colorectal cancer screening showed that the influences of socioeconomic factors and ethnicity are often not assessed; when assessed, participation gradients varied from moderate (66% vs 71%) to great (35% vs 61%) across socioeconomic and ethnic groups.25

Measures to reduce socioeconomic inequality in cancer screening

Although socioeconomic inequality in cancer screening programmes is well documented, evidence for interventions aimed at improving equity is poor. Various interventions have been studied, such as media campaigns, text message reminders, increased frequency of screening invitations and adjusted reminders to non-participants.13 Poor language skills may be a barrier, but a recent randomised trial did not show greater participation in breast cancer screening when the invitation was translated into immigrants’ original language.32 A study of healthcare interventions in Brazil suggests that socioeconomic inequality is greater in the implementation phase than when an intervention is established.33 Frequent monitoring and evaluation of participation rates in different socioeconomic groups to define determinants of low participation is considered essential.25

Unorganised PSA testing prior to the OPT

PSA testing was less common (14% vs 37%) and the gradients of PSA testing across different socioeconomic groups were smaller when the testing was unorganised compared with the subsequent organised testing programme. These results agree well with previous reports.11 12 34 35 The proportion of men with a raised PSA value was more than twice as high in the unorganised versus in the OPT setting. This suggests that many men in the pre-OPT period were PSA tested because of urinary tract symptoms, in contrast to the OPT programme to which men were actively invited, and therefore, less likely to have symptoms. Our results suggest that the influence of socioeconomic factors on obtaining medical care for symptoms differs from their influence on participation in an organised screening programme after an invitation.

Strengths, weaknesses and limitations

Strengths of our study include that the investigated OPT and pre-OPT groups were both population-based and reasonably large and that OPT is part of standard healthcare as opposed to a screening trial setting. Furthermore, the Swedish register of individual-level socioeconomic data has near-complete coverage and is highly accurate.36 A potential weakness is the 7–8 years gap between the pre-OPT and OPT settings, but there were no changes in the healthcare organisation or PSA testing policy in these years, so we do not believe that the time gap affected the investigated socioeconomic gradients.

Limitations of generalisability include that we assessed associations between socioeconomic factors and participation in the implementation phase of an organised testing programme, so they may not mirror those in a well-established programme. They also include the narrow, young age group of 50 years. The influence of socioeconomic factors on screening participation may be different in older men, who more often have lower urinary tract symptoms and experience prostate cancer among friends and family members. It may also be different in healthcare systems that rely on primary care for referral to cancer screening. Home testing of PSA, which is currently being evaluated in some Swedish OPT projects, may also affect the socioeconomic differences in participation. Moreover, a future recommendation in favour of a national screening programme would probably increase the overall participation rate and may increase participation more in some socioeconomic groups than others. Furthermore, the studied OPT period coincided with the COVID-19 pandemic, which may have affected the results. Finally, Sweden is a country with relatively narrow socioeconomic gradients,37 a public healthcare system and widespread unorganised PSA testing.18 The associations between socioeconomic factors, PSA testing and prostate cancer screening may be different in countries with wider socioeconomic gradients, an insurance-based healthcare system or less frequent PSA testing.