IJPDS International Journal of Population Data Science IJPDS 2399-4908 Swansea University 10.23889/ijpds.v9i5.2589 9:5:105 Socio-demographic variations in prostate cancer diagnosis recording: Primary care compared to the Cancer Registry in England Somathilake Gayasha 1 Ford Elizabeth 2 Armes Jo 1 Moschoyiannis Sotiris 3 Otter Sophie 4 Lemanska Agnieszka 1 5 School of Health Sciences, Faculty of Health and Medical Sciences, University of Surrey Department of Primary Care and Public Health, Brighton and Sussex Medical School Department of Computer Science, Faculty of Engineering and Physical Sciences, University of Surrey Royal Surrey County Hospital Data Sciences Department, National Physical Laboratory 18 09 2024 2024 9 5 2589 This work is licenced under a Creative Commons Attribution 4.0 International License. This article is available from the IJPDS website at: https://ijpds.org/article/view/2589 Introduction

In the UK, primary care data are often used for cancer-related research, but the accuracy of cancer information is uncertain.

 Objective

We investigated socio-demographic variation based on the recording date of prostate cancer diagnosis between primary care and the National Cancer Registry (CR).

 Approach

We utilised a data extract of 1,600,000 patients over 65 years from Clinical Practice Research Datalink (CPRD). We extracted prostate cancer diagnoses using Read and SNOMED-CT codes from primary care, and ICD-10 from CR. Initial code entry determined diagnosis dates. We categorised recording timing differences as earlier, same-day or later in primary care than CR and used the chi-squared test and logistic regression (adjusted for recording year) to compare these discrepancies across age, deprivation, and ethnicity.

 Results

We included 26,875 men with prostate cancer diagnoses commonly recorded in both sources during 2000-2016 (1,030 excluded with missing ethnicity). Compared to CR, 1,747 (7%) had diagnoses recorded on the same day in primary care, while 20,615 (77%) had later recordings with a median delay of 21 days (IQR: 13-38). Age at diagnosis was associated with recording discrepancies (p<0.001); older men were more likely to have earlier/same-day recordings. Adjusted ORs for age groups were 1.4 (95%CI: 1.3-1.5) for 60-69, 1.3 (1.2-1.5) for 70-79, and 0.9 (0.8-1.0) for ≥80 compared to <60. No associations were found between deprivation (p=0.096) or ethnicity (p=0.067) and recording differences.

 Conclusion/ Implications

The discrepancy in prostate cancer information between primary care and CR underscores potential biases in studies relying solely on one data source.