Abstract
Background The urgent-referral Cancer Patient Pathway for Non-Specific Symptoms and Signs of Cancer (NSSC-CPP) was introduced in Denmark in 2012 to reduce delays in cancer diagnoses. In Region Zealand, it includes direct referral from GPs to contrast-enhanced computed tomography of thorax, abdomen, and pelvis (ceCT-TAP). In 2013, the NSSC-CPP cancer prevalence was 20%, but easy access to computed tomography (CT) may change referral patterns.
Aim To examine cancer prevalence, diagnostic accuracy of ceCT-TAP, and changes in NSSC-CPP referral patterns.
Design and setting Retrospective cohort study conducted in Region Zealand, Denmark.
Method We included patients with non-specific symptoms or signs of cancer who were referred by GPs between 1 July and 31 December 2019. Primary endpoints were cancer prevalence and diagnostic accuracy of ceCT-TAP. Secondary endpoints were cancer types, referral trends between 2012 and 2019, and prevalence of GP-reported symptoms and clinical findings. Patients were followed up until cancer diagnosis or up to 12 months after ceCT-TAP, whichever came first.
Results In total, 729 referrals were recorded, a five-fold increase compared with 140 referrals in the same period of 2013. Malignancy was diagnosed in 95 (13%) patients. Twelve patients had a false-negative ceCT-TAP result, yielding a negative likelihood ratio (LR) for malignancy of 0.15.
Conclusion Despite a five-fold increase in GP referrals for ceCT-TAP since 2012, it remains an important diagnostic tool, identifying malignancy in one in seven patients. However, as the LR- was >0.10, a normal ceCT-TAP does not convincingly rule out cancer and should not stand alone as the only decision-support tool.
How this fits in
This retrospective study validates the efficacy of contrast-enhanced computed tomography of thorax, abdomen, and pelvis (ceCT-TAP) within Denmark’s Cancer Patient Pathway for Non-Specific Symptoms and Signs of Cancer. Despite a five-fold referral increase in GP referrals for ceCT-TAP since 2012, the scans identified malignancies in 13% of cases. This highlights the pathway’s role in facilitating earlier cancer detection; however, limitations were noted: a negative likelihood ratio of 0.15 underscores that certain malignancies remain undetected, necessitating a multimodal diagnostic approach to enhance accuracy.
Introduction
Absence of organ-specific symptoms in cancer is associated with longer time to diagnosis and treatment onset; this, in turn, is associated with advanced cancer stage at diagnosis and poorer survival rates.1–3 To address this, the Cancer Patient Pathway for Non-Specific Symptoms and Signs of Cancer (NSSC-CPP) was implemented in 2012 in Denmark. It requires GPs to follow a two-step process: step one includes taking a medical history and conducting a physical examination, and performing a standardised blood test panel is recommended; if inconclusive, step two is undertaken and a contrast-enhanced computed tomography of the thorax, abdomen, and pelvis (ceCT-TAP) is performed.
We have previously reported on the key role of ceCT-TAP in the NSSC-CCP, and found that 20% of the patients referred were diagnosed with malignancy in a 2013-2015 cohort.4 The negative likelihood ratio (LR-) of ceCT-TAP within the NSSC-CCP was 0.1 with 98% negative predictive value (NPV).4 In 2015, the National Institute for Health and Care Excellence altered the recommendation of active work-up from a 5% to a 3% risk of cancer5.. Thus, with the present study, we aimed to investigate both trends in cancer prevalence and number of referrals, as well as diagnostic accuracy of ceCT-TAP and prevalence of GP-reported symptoms and signs at referral to ceCT-TAP.
Method
Design and patient population
All citizens in Denmark have access to tax-paid, free-for-all healthcare, which is organised into five regions; Region Zealand serves a population of 835 000 inhabitants. GPs serve as primary care gatekeepers, initiating and coordinating referrals. Based on NSSC-CPP results, the GP decides on treatment and care, or referral to specialised hospital department.
This retrospective single-centre cohort study included patients aged ≥18 years, who were referred by their GP via the NSSC-CPP for ceCT-TAP at the Department of Radiology (Zealand University Hospital, Roskilde and Køge, Region Zealand, Denmark) between 1 July and 31 December 2019. Annual referral numbers for ceCT-TAP were reported from 2012 to 2019 (1 July–31 December).
Computed tomography
CeCT-TAPs were performed using multi-row detector scanners (Philips or Siemens). Intravenous contrast was administered in the portal-venous phase for the thorax and abdomen, and in the arterial phase for the liver, except in patients with an estimated glomerular filtration rate of <30, who were scanned without contrast. All scans were reviewed by a team of general radiologists.
Radiological findings from ceCT-TAPs were categorised into the following four groups based on the original descriptions provided by one of the researchers, who is an experienced oncological radiologist and was blinded to clinical data:
Group 1: no cancer and no abnormal findings;
Group 2: no suspicions of cancer but findings that, for other reasons (for example, aortic aneurysms), warrant further work-up;
Group 3: possible cancer, abnormal findings that could be malignant; and
Group 4: probable cancer.
CeCT-TAPs initially categorised as Group 1 or 2 were re-evaluated if a diagnosis of cancer was made during the study period to determine whether the cancer was identifiable in retrospect. The same oncological radiologist performed re-evaluation.
Data collection
The Danish Patient Safety Authority and the Danish Data Protection Agency approved the study before data collection was carried out.
Patients were identified using a unique code for NSSC-CPP ceCT-TAP in the Radiology Information System (RIS). No efforts were made to identify incorrectly tagged ceCT-TAPs. Patient data were retrieved from electronic patient records and national health databases, including RIS, Picture Archiving and Communication System, the Epic regional electronic health record system, and the Danish Pathology Registry (excluding non-melanoma skin cancer). The date of the ceCT-TAP was defined as day 0, and a cancer diagnosis within the subsequent 12 months was considered the diagnostic outcome. All cancers in the subsequent 12 months were retrieved from the Danish Pathology Registry.
Patient symptoms and clinical findings reported by the GP at the time of referral were extracted from the original ceCT-TAP referral texts, which were written in free-text format without the use of checkboxes.
We recorded whether patients had undergone a full blood test panel, as recommended by NSSC-CPP guidelines. This panel includes: red and white blood cell counts with differentials; platelets; renal function tests (including total calcium and albumin); liver function tests; C-reactive protein; glucose; pancreas-specific amylase; lactate dehydrogenase; myeloma protein; immunoglobulins G, A, and M; and coagulation factors II, VII, and X. As data on patients who did not proceed to ceCT-TAP within the NSSC-CPP were unavailable, we could not accurately assess the diagnostic performance of blood tests for malignancy.
Statistical analysis
Categorical variables were presented as numbers and percentages and compared using Pearson's χ² test or Fisher's exact test, as appropriate. Continuous variables were presented as means and compared using Student's t-test. Statistical significance was defined as P<0.05. To account for multiple testing, we applied a false discovery rate correction (Benjamini-Hochberg procedure) to all P-values across the 35 variables examined. A Q-value of <0.05 was considered statistically significant. A true positive ceCT-TAP was defined as a radiological classification of Groups 3 or 4 with a cancer diagnosis within the following 12 months. A true negative ceCT-TAP was defined as a radiological classification to Groups 1 or 2 and no cancer diagnosis within the following 12 months. Based on these categorisations, we calculated true positives, true negatives, false positives, false negatives, sensitivity, specificity, positive likelihood ratios (LR+s), LR-s, positive predictive values (PPVs), and NPVs.
Results
Study population
A total of 763 patients were referred for ceCT-TAP between 1 July and 31 December 2019 were identified: 34 were excluded from further analysis due to duplicate referrals, non-attendance, or cancellations (Supplementary Figure 1), thus 729 cases were eligible for inclusion. Figure 1 illustrates the annual increase ceCT-TAP as part of the NSSC-CPP, which equates to a five-fold increase from 2012.
Cancer prevalence (primary endpoint)
In total, 95 (13,0%) patients were diagnosed with cancer. CeCT-TAP’s doiagnostic rates are shown in Table 1.
Fifty-two patients died during the 12-month follow-up period, of whom 37 (71.2%) were diagnosed with cancer. A worst-case scenario — in which fatalities in patients not diagnosed with cancer were considered false-negative cancer cases — resulted in only minor changes to the diagnostic performance (Appendix Table S1
ceCT-TAP classification and post-CT referrals
Table 2 shows the radiological classification of ceCT-TAP into increasing likelihood of malignancy. Most patients were classified as having a normal CT (Group 1) and, of these, only 33% were referred for further work-up. Table 2 shows a positive and significant correlation between the severity of radiological findings (CT groups) and the likelihood of referral for further diagnostic work-up (P<0.005, χ2 test).
Table 3 shows the distribution of cancer diagnoses, and that pancreatic cancer (17,0%), lung cancer (14,0%), and haematological malignancy (13,0%) were the most predominant. Table 3 also shows that 12 patients in Groups 1 and 2 (no suspicion of cancer) were diagnosed with malignancy in the study period. Following completion of the 12-month follow-up, one case of lung cancer and one case of oesophageal cancer were reclassified as probable cancer (Group 3); the remaining 10 cancers — haematological cancer (n = 6), rectal cancer (n = 1), cerebral glioblastoma (n = 1), breast cancer (n = 1), and malignant melanoma (n = 1) — were not visible in retrospect. Only 50,6% of all refered patients had the mandatory blood test panel prior to ceCT-TAP.
Table 4 shows that patients diagnosed with cancer were older and had significantly higher 12-month mortality than those not diagnosed with cancer. It also depicts symptoms and signs reported by the GP at the point of referral for ceCT-TAP. In 138 (19%) cases GPs reported having a ‘gut feeling’ that cancer was present; gut feeling was one of few symptoms/signs that was associated with malignancy (Table 4). After correction for multiple comparisons, only patient’s age and reported anaemia or a GP’s ‘gut feeling’ remained statistically significant (data not shown). The LR+s and LR-s for GP ‘gut feeling’ were 1.7 and 0.9 respectively (data not shown).
Discussion
Summary
This study shows that, despite a fivefold increase in GP referrals for ceCT-TAP in the NSSC-CPP from 2012 to 2019 due to non-specific symptoms and signs, the prevalence of detected malignancy declined from 20% to 13%. This corresponded to an increase in numbers-needed-to-scan from five to seven. The absolute number of cancers detected increased during the study period.
Diagnostic accuracy analysis showed that ceCT-TAP alone could neither confidently rule in cancer (LR+ <10) nor rule it out (LR− >0.1). Not all cancer types were readily detectable on ceCT-TAP, as lesions may be located in non-scanned regions or be radiologically occult. This was observed in 10 of the 12 missed cancers in our cohort.
Symptoms reported by GPs were heterogeneous, with weight loss being the most frequently reported symptom.
Among all patients undergoing ceCT-TAP, 58.2% had no abnormal findings, and 67.2% of these patients had no further hospital follow-up. In 27% of patients, ceCT-TAP findings prompted referral to non-cancer pathways. Overall, one in seven referred patients was diagnosed with cancer.
Strengths and limitations
Strengths of this study include a clearly defined cohort of consecutive patients within a predefined timeframe, minimal missing data, complete follow-up, and access to regional electronic patient records and national pathology data. All patients were identified through procedure codes, and none were lost to follow-up. Furthermore, diagnostic accuracy was evaluated using established measures, including likelihood ratios (LR+ and LR−), allowing assessment of the clinical utility of ceCT-TAP beyond conventional measures such as sensitivity and specificity.6
Several limitations should be considered. First, symptom data were based on information provided by GPs in ceCT-TAP referral forms and may therefore differ from patient-reported symptoms.7–9 Second, lack of access to GP medical records means that additional symptoms or clinical findings may not have been documented in the referral text. Finally, Referral bias may also have influenced which symptoms were reported, as GPs might prioritise those likely to result in accepted referrals.
Comparison with existing literature
Several studies have reported declining cancer prevalence among patients referred through NSSC pathways as familiarity with referral systems has increased over time10–13 . Similarly, we observed a decline in malignancy prevalence from 20% in 2013 to 13% in 2019 despite a substantial increase in referral activity. Grønnemose et al10, studying another Danish region in which GPs cannot directly refer to ceCT-TAP, observed a more marked drop in malignancy prevalence from 40% in 2014 to 19% in 2021. In that cohort, 75% of patients were referred by their GP to the hospital-based diagnostic centre, corresponding to approximately 14% of the overall cohort being diagnosed with cancer following GP referral, which is similar to the 13% observed in the present study. Differences between studies may reflect regional organisation of diagnostic work-up, as the diagnostic centre described by Grønnemose et al. also received referrals from hospital departments, unlike our direct GP referral system.
The distribution of cancer types was broadly consistent with previous evaluations of the NSSC-CPP pathway. Pancreatic and lung cancers remained among the most frequently diagnosed malignancies, both of which are known to present with vague or non-specific symptoms.14 Haematological cancers were as prevalent as lung cancer at 12-month follow-up, but only 50% were visible on the initial ceCT-TAP, reflecting the limitations of this modality for early detection.
Diagnostic accuracy of ceCT-TAP alone remains insufficient to guide further work-up, as the LR+ was below 10 and the LR− was above 0.1, thresholds generally accepted as providing strong diagnostic evidence for ruling in or ruling out disease, respectively.6
Diagnostic accuracy findings were also consistent with previous studies demonstrating limitations of ceCT-TAP as a stand-alone diagnostic test in patients with non-specific symptoms. PET-CT has shown promising results in the NSSC-CPP setting, with significantly higher specificity and diagnostic accuracy, together with a non-significant trend towards higher sensitivity compared with ceCT-TAP.15
Finally, the association between GP ‘gut feeling’ and subsequent cancer diagnosis is consistent with previous studies. 16–18 However, it cannot independently confirm or exclude cancer. In contrast symptoms, such as weight loss, were not consistently associated with cancer; this aligned with findings by Ingemann et al,14 but differed from those of de Chiffre et al.19
Implications for research and/or practice
The findings support the continued use of ceCT-TAP as part of the diagnostic work-up for patients referred to the NSSC-CPP with non-specific symptoms suggestive of cancer. However, imaging findings should be interpreted in the context of the overall clinical assessment, and persistent or progressive symptoms should prompt further evaluation when clinically indicated.
Future research should focus on optimising diagnostic pathways for patients presenting with non-specific symptoms, including evaluation of alternative or complementary imaging strategies and their impact on diagnostic accuracy, resource utilisation, and patient outcomes. Further studies are also needed to examine patient trajectories following negative ceCT-TAP findings and to identify factors associated with subsequent cancer diagnoses. Such knowledge may help refine follow-up strategies and improve the efficiency of NSSC-CPP pathways.
Notes
Funding
The authors have not declared a specific grant for this research from any funding agency in the public, commercial, or not-for-profit sector.
Ethical approval
The study was approved by the Danish Patient Safety Authority and the Danish Data Protection Agency before data collection was initiated.
Provenance
Freely submitted; externally peer reviewed
Data
Data are available from the corresponding author, Mads Rømer Skøtt, upon reasonable request.
Competing interests
The authors declare that no competing interests exist.
- Received April 9, 2025.
- Revision received August 21, 2025.
- Accepted October 1, 2025.
- Copyright © 2026, The Authors
This article is Open Access: CC BY license (https://creativecommons.org/licenses/by/4.0/)
References
LinkedIn
Jump to section
Keywords
More in this TOC Section
Related Articles
Cited By...
