GP referrals for suspected cancer in Ireland: protocol for a cross-sectional study (GRACCHUS)

Cancer burden and the importance of early diagnosis

Cancer remains a leading cause of mortality worldwide, accounting for one in six deaths¹. In Ireland, approximately 43,000 people are diagnosed with cancer each year, with over 9,000 cancer-related deaths, representing nearly one-third of all national mortality². Lung cancer is the most common cause of cancer death in Ireland, responsible for 20% of cancer-related mortality, followed by breast cancer in women, prostate cancer in men, and colorectal cancer in both sexes².

Cancer survival is influenced by multiple factors, including tumour biology, patient comorbidities, and access to effective treatments. However, for most tumour types, stage at diagnosis is the strongest predictor of survival³. In Ireland, 30% of lung cancers are diagnosed at stage I or II, compared to 41% of colorectal cancers, 65% of prostate cancers and 76% of breast cancers. Five-year net survival for lung cancer is just 24%, compared to 66% for colorectal cancer, 93% for prostate cancers and 88% for breast cancer^4,5. Given that early-stage cancers are more amenable to less complex and curative treatment, strategies that facilitate timely diagnosis are critical for improving patient outcomes and reducing healthcare and treatment burden^3,6,7.

Rapid Access Clinics in Ireland: structure, utilisation, and limitations

The National Cancer Control Programme (NCCP) was established in 2007 as a directorate of the Health Service Executive (HSE) to coordinate cancer services, standardise care, and improve survival rates. A key initiative has been the development of Rapid Access Clinics (RACs), specialist-led services designed to expedite diagnosis and specialist assessment for suspected breast, lung, prostate cancer, and melanoma—the four most commonly diagnosed cancers in Ireland⁸.

RACs operate within designated cancer centres, with defined targets for appointment scheduling, aiming to ensure timely evaluation following GP referral⁵. International evidence suggests that structured pathways such as RACs improve access to diagnostic services, reduce diagnostic delays, and increase the proportion of cancers diagnosed at an early stage^6,9. Despite this, Irish data suggest that diagnostic pathways may not be fully optimised, as a significant proportion of cancers continue to be diagnosed at an advanced stage⁵.

Variation in GP referral behaviour and patient outcomes

General practitioners (GPs) are central to early cancer detection, with the majority of symptomatic cancer patients initially presenting in primary care⁹. However, referral pathways are complex, and the decision to refer is influenced by clinical guidelines, GP experience, patient characteristics, and healthcare system constraints^10,11. While RACs provide an expedited pathway for specialist assessment, referral decisions are complex, and substantial variation in GP referral behaviour has been documented in international studies¹².

In Denmark, where structured urgent referral pathways have been in place since 2008, approximately 27% of lung cancer diagnoses occur via fast-track pathways⁹. However, GP referral rates for suspected cancer vary significantly between practices, even after adjusting for case-mix, practice size, and patient demographics^12–14. Notably, higher referral rates are associated with lower emergency presentation rates and improved cancer survival¹⁵. These findings suggest that greater utilisation of structured referral pathways may contribute to earlier cancer detection and better patient outcomes.

In contrast, there is limited evidence on the extent of referral variation within Ireland and its impact on cancer diagnosis and outcomes. The "Cancer Care in Ireland 2020" report recorded 44,245 new attendances at RACs for breast, prostate, and lung cancers in a single year¹⁶. However, the conversion rate (proportion of referrals resulting in a cancer diagnosis) remains unclear, making it difficult to assess the efficiency of the RAC system. Additionally, how referral rates vary by patient demographics, GP practice characteristics, and geographic location has not been explored, limiting opportunities for targeted improvements in diagnostic pathways.

Given the established link between referral patterns and cancer outcomes^6,15, a detailed examination of RAC referral trends is necessary to identify potential barriers to timely cancer diagnosis. These may include regional disparities in access to diagnostic services, differences in GP decision-making and adherence to referral guidelines, and variations in patient presentation, such as symptom severity and comorbidities. Furthermore, analysing the relationship between referral rates and conversion rates will provide a clearer picture of how effectively RAC pathways function in practice. Identifying patterns of under-referral or over-referral will be essential for improving diagnostic efficiency, ensuring that referral decisions align with clinical need. This evidence will inform GP education and the refinement of clinical guidelines to promote appropriate referrals, support policy interventions aimed at addressing gaps in cancer diagnostic services, and enhance quality improvement initiatives to optimise the equitable and efficient use of RACs.

This study is the first in Ireland to propose a comprehensive, retrospective cross-sectional analysis of RAC referrals, using electronic health records (EHRs) from general practice. By quantifying referral trends, identifying predictive factors, and assessing diagnostic outcomes, this research will generate critical evidence to strengthen RAC pathways and improve early cancer detection.

Aims and objectives

This study aims to examine general practice referrals to RACs for suspected cancer in Ireland, evaluating referral patterns, variation across practices, and the impact on cancer diagnosis and outcomes. Specific objectives include:

•   To characterise RAC referral patterns by cancer type, volume, and inter-practice variability over time.

•   To examine demographic, geographic, and clinical factors influencing referral rates.

•   To determine the diagnostic conversion rate, time from referral to diagnosis, and stage at diagnosis for patients referred to RACs.

•   To assess cancer treatment and survival outcomes following RAC referral.

Study design

This is a retrospective study that will analyse electronic health record (EHR) data from Centric Health’s network of general practices in Ireland. Specifically, we will examine electronic rapid access cancer referrals for breast, prostate, lung cancer, and melanoma, by GPs in Ireland from 2013–2024. The data collection and analysis for the study will be conducted throughout 2025. The study will be conducted in two phases.

The first phase has three steps: 1) a descriptive analysis of utilisation of RAC pathways over time and how this varies by GP practice; 2) a series of descriptive analyses, which, for each referral type, examines how the prevalence or degree of a referral characteristic varies over time, and 3) a series of negative binomial regression analyses which seek to explain the variation in the number of referrals (of a particular type) from a GP practice varies as a function of GP practice characteristics, patients characteristics and time.

The second phase involves: 1) a descriptive analysis of outcomes of RAC referrals over time, and 2) an exploration and analysis of characteristics predictive of cancer diagnosis and outcomes over time. We will adhere to the SPIROS (Standardized Protocol Items Recommendations for Observational Studies) and STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) Checklist for reporting protocols and cross-sectional studies, respectively^17,18.

Study setting & participants

The setting includes GP practices from the Centric Health practice network. Eligible participating practices are GP practices utilising the Socrates patient management system (Socrates Healthcare Ltd. Sligo, Ireland; https://www.socrates.ie/). Participation will be discussed with practices from the targeted networks after outlining the study's scope and significance. These inclusion criteria are designed to ensure a reliable dataset and consistency in data collection methods.

Sample size

As this is an observational study of a general practice population that aims to characterise current practice, and generate hypotheses for further investigation, there is no intent to conduct inferential analyses beyond the studied population. As such, a sample size calculation is not necessary.

Variables

The primary outcome variable for phase 1 is:

The primary outcome variable for phase 2 is:

Data collection and management

Data extraction tool

A data extraction tool, based on the Visual Basic for Applications (VBA) language was designed for phase 1 of this study (Microsoft. (2025). Microsoft Office [Visual Basic for Applications]. Available from https://www.microsoft.com/en-us/microsoft-365). It was designed to extract data in Extensible Markup Language, which is the output produced for referrals in some EHRs in Ireland. It then converts this data into a format suitable for tabulation and analysis in Microsoft Excel. It will collect variables as outlined in Table 1. The tool will be refined through a trial run of pilot practices with iterative debugging. Validity of the data extracted will be determined by comparing the proportion of records extracted relative to the known number of records. A successful extraction rate, indicating adequate completeness of data, namely complete extraction of information from a referral file, will be defined as exceeding a threshold of 95%.

Electronic Health Record chart review

Patient EHRs will be manually searched by the study team and medical students under the supervision of Centric Health in each practice to extract the data for phase 2, specifically extraction the information listed in Table 1. The list of patient charts to review will be produced from the phase 1 data extraction process. Researchers will login to the EHR using a specific temporary login provided by the Practice Manager. This will have an associated audit log providing oversight of the data viewed during the chart review process. A Standard Operating Procedure (SOP) for the data collection process will be developed and used to train the relevant personnel, ensuring consistency and quality of data throughout phase 2. Sample data collection spreadsheets will be freely available via Open Science Framework.

Data governance

Data will be pseudonymised following extraction from the clinical records in Phase 1, ensuring the removal of identifiable personal information. The pseudonymised data will remain on the same secure server in a password-protected document with access strictly limited to authorised members of the research team. The unique identifier linking this data to identifiable patient information will be stored separately in a secure, password-protected file.

Bias

Potential sources of bias include recruitment from a specific network of practices, and exclusion of non-Centric health practices. These will be addressed and quantified by ensuring a range of practices are included and by analysing the representativeness of participating practices. Efforts to mitigate bias will be documented and discussed in the study's findings.

Ethics

Ethical approval for the study was granted by the Irish College of General Practitioners Research Ethics Committee (REC) on 12th December 2024 (reference no. 2662). According to the Amendments (January 2021) to the Health Research Regulations 2018, individual consent is not required for a low risk, retrospective chart review that has gained approval by the appropriate REC and is conducted by authorised parties¹⁷. There is low risk of potential harm to participants of the study. There is no intent to involve the patients or the public in the design of this study protocol.

Statistical approach

Data analysis for phase 1 of the study consists of three parts focusing on GP practice characteristics, referral volume trends and clinical information trends respectively.

Data analysis for phase 2 of the study will focus on conversion rates for referrals as well as cancer stage at diagnosis and time from referral to diagnosis. First, conversion rates for all referrals over the study period will be presented by cancer type. Then rates will additionally be stratified by year of referral. The timeline from referral to diagnosis including pertinent investigations will be described for each cancer type for each practice. Finally, conversion rates will be characterised by all of cancer type, year of referral, and GP practice. The (TNM and/or Gleason) stage of cancer at diagnosis will be summarised for each cancer type.

Implications for practice and policy

The anticipated results of this study are expected to provide a foundation for future research and inform policy and practice surrounding cancer diagnosis – particularly in primary care - in Ireland. We expect that the insights gained will be crucial in guiding population health policy, particularly in improving early detection strategies for cancer. By understanding referral patterns and associated cancer outcomes, participating GP practices can benchmark their behaviours against wider trends, potentially leading to quality improvement initiatives or an informed audit cycle.

Limitations

While aiming for comprehensive and representative findings, we acknowledge potential limitations in our study. These may include the representativeness of the sample and inherent biases in referral practices. Moreover, the reliance on retrospective data from practices might introduce variability in the data quality. We plan to address these limitations through rigorous study design and transparent reporting of our methods and findings. Finally, while rapid access pathways are viewed as the primary route for referring patients with these suspected cancers, there are other potential routes of referral (paper-based or private referral pathways) that may not be captured in this dataset. However, the focus of this analysis is on the use of rapid access pathways rather than other routes to diagnosis.

Dissemination

We plan a comprehensive dissemination strategy, targeting both academic and healthcare professional audiences. Findings will be submitted to peer-reviewed journals, presented at relevant conferences, and shared with healthcare policy makers. Additionally, we aim to engage with the wider public through media releases and open-access publications to ensure the study's implications are communicated broadly.

Artificial Intelligence

Artificial intelligence (AI) assisted technologies have not been used in the preparation of this protocol and there is no intent to use them in the execution of this study.