Research

The sensitivity of decision support tools for identifying patients with pancreatic cancer: an observational study

Rachel E Neale, Susan J Jordan, Bridie Thompson, Christina M Bernardes, Judi Adams, Christopher Baggoley, Savio George Barreto, Catherine M Baxter, Daniel Croagh, Benedict Devereaux, Jon Emery, Louisa G Collins, Rajit Gilhotra, Paul Grogan, Luke Hourigan, Javiera Martinez-Gutierrez, Andrew J Metz, Stephen Philcox, Meena Rafiq, Joel Rhee, Silja Schrader, Michelle Stewart, John Windsor, John Zalcberg and Mary Waterhouse
BJGP Open 30 June 2026; BJGPO.2025.0142. DOI: https://doi.org/10.3399/BJGPO.2025.0142

Abstract

Background Pancreatic cancer causes non-specific symptoms, potentially leading to delays in diagnosis. Decision support tools may help primary care practitioners to triage patients for pancreatic imaging.

Aim To investigate the sensitivity of three different tools for identifying patients who may have pancreatic cancer.

Design & setting An observational study in Australia.

Method We investigated the performance of the Risk Assessment Tool (RAT) for pancreatic cancer, the QCancer® tool, and a tool developed through a consensus process led by QIMR Berghofer (the QPaC Tool). We applied these tools to people with pancreatic cancer who were interviewed about their symptoms on first presentation to a clinician. We designated patients as ‘flagged’ by each tool if they met specific criteria, and calculated the percentage flagged (that is, the sensitivity). Participants with jaundice were excluded from analyses of QCancer®.

Results We included 190 participants in analyses of the RAT and QPaC Tool (142 in analyses of QCancer®). The sensitivity of the QPaC Tool and the RAT were 54% and 27%, respectively. QCancer® had a sensitivity of 14%, at a probability threshold of 1%; in the same 142 participants, QPaC and the RAT flagged 44% and 7%, respectively.

Conclusion The QPaC Tool was the most sensitive, largely owing to its inclusion of severe epigastric pain and emphasis on diabetes, but it has unknown specificity. More research is needed to determine whether any tool could reduce delays in diagnosis; in the interim, the QPaC Tool may support clinicians to consider pancreatic cancer in their differential diagnoses.

How this fits in

Several decision support tools designed to reduce delays in diagnosis of pancreatic cancer have been developed. The sensitivity of these has not been previously compared. Only one of the tools (the QPaC Tool) had a sensitivity of more than 50%, but its specificity is unknown and it may be difficult to embed into GP software owing to nuances in symptoms. Nevertheless, it might be useful as a prompt to encourage GPs to consider pancreatic cancer on a list of differential diagnoses.

Introduction

Pancreatic cancer is the seventh most commonly diagnosed cancer in developed countries, but the third most common cause of cancer death. This difference between incidence and mortality is underpinned by low survival,1 which is primarily owing to the late stage at which most pancreatic cancers are diagnosed.

Reducing delays in diagnosis and treatment initiation may improve clinical outcomes such as stage distribution, resectability, and overall survival.2 Patient experience of the care process is also an important consideration, as perceived diagnostic delay can lead to higher cancer-related distress.3

Delayed diagnosis in patients with pancreatic cancer is common. One Australian study found that more patients with pancreatic cancer than with any other cancer type reported three or more GP visits before diagnosis.4 In a UK study, 41% of patients with pancreatic cancer had three or more GP visits before being referred to a specialist, the highest of any cancer type other than multiple myeloma.5 In the UK, the median health-system interval was 71 days; for 25% of patients it was more than 160 days.6

Pancreatic cancer frequently presents with non-specific symptoms, and it is challenging for GPs to determine when these may be indicators of pancreatic cancer, rather than of a more common and less serious condition. Several decision support tools have been developed to aid in early diagnosis of pancreatic cancer in primary care. The Risk Assessment Tool (RAT) for pancreatic cancer consists of a matrix table of positive predictive values (PPV) for individual symptoms and pairs of symptoms.7 QCancer® calculates the probability of cancer (any cancer and specific cancer types) based on demographic and clinical risk factors, and symptoms.8 A decision guide, developed by a consensus process led by QIMR Berghofer, (hereafter called the QPaC Tool) has a qualitative approach. It divides symptoms into tiers, according to the urgency of investigation, with tier 2 allowing for more common conditions to be ruled out before proceeding to investigations of the pancreas.9

We aimed to describe the sensitivity of the RAT, QCancer®, and QPaC tools in a cohort of patients with pancreatic cancer who retrospectively self-reported the signs, symptoms, and risk factors that were present when they first presented to a clinician.

Method

Participants

The data used were collected for an Australian study focused on understanding the extent, causes, and consequences of diagnostic delay: the Pathways Study. We recruited participants between November 2021 and December 2023 as follows: (i) the Queensland Cancer Registry wrote to people who had been diagnosed with pancreatic cancer during the previous 6 months; (ii) clinicians treating patients at several public and private clinics in two Australian states (Queensland and Victoria) asked patients for permission to release their details to the Pathways Study; (iii) PanKind, a non-government philanthropic organisation, promoted the study via email lists and social media; (iv) participants of another research study, which was recruiting via additional hospitals in Queensland, were invited to register for the Pathways Study.

We excluded participants if their pancreatic cancer was diagnosed>2 years before study consent, or was identified incidentally or during routine surveillance. The QCancer® tool does not consider jaundice. Hence, when evaluating QCancer®, we excluded participants with jaundice and those for whom it was not possible to calculate a QCancer® probability, owing to missing data.

Recording and deriving indicators used in the decision support tools

Pathways Study participants completed a survey about demographic and lifestyle characteristics and their medical history before being diagnosed with pancreatic cancer. They then took part in a semi-structured interview about their symptoms and diagnostic journey. All data were self-reported.

Symptoms

During the interview, we prompted participants to recall whether they experienced any of the following symptoms on first presentation to a healthcare provider: abdominal or back pain (including location and nature); weight loss (including amount and time frame); fatigue; nausea; vomiting; changed bowel habits; loss of appetite; and changed food or drink preferences. Participants were not explicitly asked about symptoms associated with cancers other than pancreatic cancer (for example, nocturia, breast lump). We did not specifically ask which symptoms were disclosed to a clinician.

The signs and symptoms at first presentation for which we derived indicators were: severe epigastric pain; mild epigastric pain; non-specific abdominal pain; jaundice; steatorrhoea; weight loss; upper back pain; any back pain; nausea; changed bowel habits; indigestion or pain after eating; anorexia or loss of appetite; new-onset diabetes; newly unstable diabetes; bloating or abdominal distension; malaise; vomiting; change in dietary preferences; pancreatitis; abdominal discomfort; and fatigue.

We assumed a participant had jaundice if they reported jaundice, pruritus, and/or dark urine. We considered a participant to have newly unstable diabetes if they had existing diabetes and they reported changes in control of their blood sugar levels in the 2 years before being diagnosed with pancreatic cancer. It was not possible to derive an indicator of biliary-type pain (included in the QPaC Tool) owing to lack of information regarding upper right quadrant pain.

Implementing the decision support tools

To evaluate the tools, we classified each participant as being flagged for suspected pancreatic cancer or not. Participants flagged by the QPaC tool were classified as being in either tier 1 (urgent investigation), or tier 2 (non-urgent: trial of management or rule out other conditions) (Table 1).

View this table:
Table 1. Criteria used to classify participants as being as being flagged for pancreatic cancer, according to the QPaC Tool and the RAT for pancreatic cancer

The criteria for the RAT used to flag participants were chosen on the basis of the National Institute for Health and Care Excellence (NICE) guidelines for urgent pancreatic cancer referral (Table 1);10 these corresponded to PPVs ≥1.5% in the RAT matrix.

We translated the QCancer® risk equations into the SAS language (SAS version 9.4), and used them to estimate probabilities of undiagnosed cancers for each participant. Details of QCancer®’s implementation are shown in Supplementary Table S1. We used multiple probability thresholds (0.25% to 3% in increments of 0.25%) to convert predicted probabilities into binary classifications. A participant was flagged for a particular cancer type if QCancer® estimated that their probability of having undiagnosed cancer of that type exceeded the threshold. For comparative analyses with the QPaC and RAT tools, we used a probability threshold of >1% because previous research demonstrated that a threshold of 0.9% achieved a PPV of 1.4%11 (similar to that used for the RAT).

Results

In total, 313 people registered interest in taking part in the Pathways Study and 232 were interviewed. We excluded 42 participants who were diagnosed >2 years before consent, or where the diagnosis occurred incidentally or through routine surveillance, resulting in 190 participants. The evaluation of QCancer® was restricted to 142 participants, after excluding 41 with jaundice, and seven with unknown body mass index.

Selected characteristics of participants are shown in Table 2. Briefly, 54% of participants were male and the mean age at first presentation was 66 years. Approximately one-quarter of participants reported a history of diabetes. At the time of interview, 29% of patients had undergone a resection of the primary tumour (Supplementary Table S2).

View this table:
Table 2. Participant characteristics

Sixty-two percent of participants experienced at least one pain symptom at first presentation. Other common signs and symptoms were jaundice (22%), changed bowel habits (18%), weight loss (17%), and malaise (16%) (Table 3). The majority (75%) of participants reported more than one sign or symptom at first presentation (Supplementary Table S3).

View this table:
Table 3. Prevalence of signs, symptoms, and diseases at first presentation, overall and stratified by sex

Overall, the QPaC Tool flagged 54% of participants, with 46% flagged for urgent investigation (Figure 1, Table 4). The RAT flagged 27% of participants. Compared with men, a greater percentage of women were missed by the QPaC Tool (Table 4).

Percentage of participants flagged for pancreatic cancer by the QPaC Tool and the RAT for pancreatic cancer. RAT = Risk Assessment Tool
Figure 1. Percentage of participants flagged for pancreatic cancer by the QPaC Tool and the RAT for pancreatic cancer. RAT = Risk Assessment Tool
View this table:
Table 4. Classification of participants according to the QPaC Tool, the RAT for pancreatic cancer, and the QCancer® tool

Among participants included in the evaluation of QCancer®, the choice of probability threshold had a marked effect on the percentage of participants flagged for pancreatic cancer (Supplementary Table S4). As the threshold increased, the percentage flagged decreased in an approximately exponential manner (Supplementary Figure S1). A similar decreasing trend was seen in the percentage flagged for at least one cancer type (Supplementary Table S4, Supplementary Figure S1).

For the comparative analysis in which we applied a >1% probability threshold, 51% of participants had at least one cancer type flagged by the QCancer® tool (69% of males, 31% of females) (Supplementary Table S4). Among males, the specific cancer type most frequently flagged was colorectal cancer (53%; 34% of these were also flagged for pancreatic cancer); in females, the most commonly flagged cancer was ovarian cancer (20%; 15% of these were also flagged for pancreatic cancer) (Supplementary Table S5). Figure 2 shows the distribution of the predicted probability of undiagnosed pancreatic cancer in this cohort, according to the QCancer® tool. Fourteen per cent of participants (21% of males; 6% of females) were flagged (Table 4). In the same cohort of participants, the QPaC Tool and the RAT identified 44% and 7%, respectively (Table 4).

Distribution of probabilities of undiagnosed pancreatic cancer in Pathways participants, according to the QCancer® tool among participants without jaundice, and with known body mass index (n = 142)
Figure 2. Distribution of probabilities of undiagnosed pancreatic cancer in Pathways participants, according to the QCancer® tool among participants without jaundice, and with known body mass index (n = 142)

The signs, symptoms, diseases, and risk factors of participants who were not flagged by each tool are shown in Supplementary Table S6. Seven participants were not flagged by the RAT despite having relevant symptoms because they were aged <60 years. The higher sensitivity of the QPaC Tool was primarily owing to the inclusion of severe epigastric pain and newly unstable diabetes. Back or abdominal pain occurred in more than one-fifth of participants missed by all tools.

Discussion

Summary

Most Pathways Study participants presented with non-specific symptoms, reflected in the suboptimal sensitivity of all three tools. Both QCancer® and the RAT failed to flag the majority of patients with pancreatic cancer at first presentation. The QPaC Tool performed better, but 46% of patients would still have been missed. This highlights the challenges in decreasing delays in the diagnosis of pancreatic cancer.

Strengths and limitations

The main strength of this study is the quality of the data underpinning the analyses. Rather than relying on broad symptom categories, we captured more detailed information; for example, we distinguished between general abdominal pain and severe epigastric pain. This level of detail was especially valuable when evaluating the QPaC Tool, which incorporates more nuanced symptom characteristics, allowing for a more accurate assessment of its clinical utility.

The Pathways Study dataset has some limitations. There is no control group, precluding calculation of specificity and predictive values. The symptoms observed may not represent the broader patient population, since a higher proportion had resectable disease than is usually observed.12 We relied on self-report of symptoms reported retrospectively, which may have led to inaccuracies. This is likely to have been a particular problem for QCancer® and QPaC, which incorporated more non-specific symptoms, with the potential that sensitivity would have been overestimated. Finally, we did not capture information about all symptoms included in QCancer®; while most were not relevant to pancreatic cancer, the performance of QCancer® may have been underestimated.

Comparison with existing literature

The three tools have different features. QCancer® simultaneously models the risk of various cancers. It presents multiple probabilities, ranked in order from the most to the least likely. While this has merit in principle, in one simulation study GPs expressed reservations about its use.13 In particular, differences in interpretation of symptoms led to variation in the risk assessments, and GPs found it difficult to reconcile the numerical output of the tool with their clinical judgement when these differed. Further, there is no clear guidance on the probability threshold that should be used to decide whether to initiate investigations, and whether this should vary by cancer type. Pathways Study participants were more likely to have been flagged with colorectal (male) or ovarian (female) cancer than with pancreatic cancer. Thus the initial imaging investigations recommended would have been colonoscopy or ultrasound, potentially leading to delay in referring a patient for a computed tomography (CT) scan of the pancreas.

The RAT presents PPVs for symptoms and combinations, and is intended to be integrated into GP software. In the UK, where the tool was developed, NICE advises that patients aged≥40 years should be referred on a suspected cancer pathway if they have jaundice; an urgent direct access CT should be considered for patients≥60 years with weight loss accompanied by abdominal symptoms or new-onset diabetes.10 Flagging patients with jaundice is arguably of limited utility, as jaundice should automatically lead to referral or urgent investigations without the need for a tool. Unexpected weight loss has been associated with 10 different types of cancer.14 While most GPs would recognise this as a red flag, red flags do not always trigger a timely referral.15 Importantly, weight loss is frequently under-reported,16,17 so it is necessary to increase reporting alongside providing guidance about how to proceed once it has been detected.

The QPaC Tool was developed based on consensus and does not present probabilities. Tier 1 aims to identify patients requiring urgent investigation of the pancreas. Tier 2 encourages clinicians to add pancreatic cancer to their list of differential diagnoses. An advantage of QPaC is that it captures specific symptom details, such as the nature of epigastric pain. This nuance was not possible in the development of the RAT or QCancer® tools, owing to their reliance on routinely collected coded electronic health record data. The specificity and predictive values of the detailed symptoms are unknown and would be challenging to accurately estimate, given the limitations of large administrative datasets. Importantly, the QPaC Tool would be more difficult than the RAT to integrate into software without changes in the way in which signs and symptoms are captured.

The QPaC tool places more emphasis than the other tools on new-onset or unstable diabetes. Urgent investigations are recommended in patients who also have other important risk factors or symptoms. Otherwise, clinicians are advised to consider a CT of the pancreas if the patient does not exhibit adequate glycaemic control in response to initial medical therapy. While there are strong associations between diabetes and pancreatic cancer,18,19 and deterioration of glycaemic control can occur up to 3 years before pancreatic cancer is diagnosed,20,21 the predictive values of the sign and symptom combinations in the QPaC Tool are unclear. Further, there is no randomised controlled trial evidence that early imaging of patients with new-onset or unstable diabetes would reduce pancreatic cancer mortality, and GPs have expressed some concern about acting on the advice owing to the risk of subjecting patients to unnecessary investigations with consequent costs and potential harms.22 More research is needed to determine whether there is a net benefit of early investigations for pancreatic cancer in people with new-onset or newly unstable diabetes.

Implications for research and practice

There is some evidence that incorporating clinical decision tools into general practice software can reduce time to diagnosis and improve the quality of referrals, decision making, or clinical outcomes,23 although there have been no studies specifically relating to pancreatic cancer. However, there are barriers to the implementation of such tools, such as disruption to workflow, lack of trust in the recommendations, and ‘flag fatigue’. All three tools we investigated have limitations, and more research is needed to determine if implementation of any would reduce pancreatic cancer mortality and/or lead to over-investigation of common symptoms. In the interim, the QPaC Tool may have some value in educational settings or as a prompt to GPs to consider pancreatic cancer in patients with non-specific symptoms.

Notes

Funding

The collection of the Pathways Study data was funded by grants from Pankind (the Australian Pancreatic Cancer Foundation), Viatris, and Perpetual Trustees IMPACT Philanthropy Grant. Cancer Australia funded the development of clinical guidance, including the analysis presented here, as part of progress to implement priorities under the National Pancreatic Cancer Roadmap (the Roadmap). The Roadmap, developed by Cancer Australia, sets out the need to improve identification of people at high risk of pancreatic cancer for targeted surveillance and improve primary health professional recognition of signs and symptoms of pancreatic cancer.

Ethical approval

The study was approved by the Royal Brisbane & Women's Hospital Human Research Ethics Committee (HREC/2021/QRBW/79630) and the QIMR Berghofer Human Research Ethics Committee (P3758).

Provenance

Freely submitted; externally peer reviewed.

Data

The data analyzed in this study are not publicly available because participants did not consent to their data being made available in this way. Researchers interested in collaboration may request access from the corresponding author, subject to a data transfer agreement.

Patient consent

All participants gave informed consent.

Acknowledgements

We would like to thank the participants who took part in the Pathways Study, and acknowledge the work of Andrea McMurtrie, Hanna Beebe, and Gretel Whiteman who interviewed participants and extracted relevant data. Recruitment was supported by the Queensland Cancer Registry with approval of the Queensland Public Health Act 2005.

Competing interests

The following authors were either directly involved in the development of the QPaC Tool, or are affiliated with QIMR Berghofer, which led the development: Rachel Neale, Stephen Philcox, Benedict Devereaux, Andrew Metz, Daniel Croagh, John Windsor, Bridie Thompson, Joel Rhee, Christina Bernardes, Catherine Baxter, Louisa Collins, and Mary Waterhouse. Andrew Metz sits on two advisory boards for Creon. Jon Emery is a member of the Clinical Advisory Board of Rhythm Biosciences. John Zalcberg declares: leadership of ICON Group, Lipotek, Praxis; Stock interests in Biomarin, Ophthea, Amarin, Frequency Therapeutics, Gilead, UniQure, Orphazyme, Moderna Therapeutics, Novava, CSL Limited, Korro; Consulting/advisory roles with Merck Sharp & Dohme, Deciphera, Revolution Medicine, FivePHusion, Genorbio, 1Global, Alloplex Biotherapeutics Inc, Oncology Republic, Duo Oncology, Taiho Oncology, Takeda, Avance Clinical, BioNTech SE, BioIntelect; Research funding from Bristol-Myers Squibb, AstraZeneca, Pfizer, IQvia, Mylan, Ipsen, Eisai, Medtronic, MSD Oncology, Servier, Astellas Pharma, Taiho Oncology; Travel support from MSD Oncology, ICON Group, Praxis. Ben Devereaux has a consultancy agreement with Olympus Australia.

  • Received July 21, 2025.
  • Accepted September 4, 2025.

This article is Open Access: CC BY license (https://creativecommons.org/licenses/by/4.0/)

References

  1. 1.↵
    1. Wang H,
    2. Liu J,
    3. Xia G,
    4. et al.
    (2020) Survival of pancreatic cancer patients is negatively correlated with age at diagnosis: a population-based retrospective study. Sci Rep 10(1):7048, doi:10.1038/s41598-020-64068-3.
    OpenUrlCrossRefPubMed
  2. 2.↵
    1. Lukács G,
    2. Kovács Á,
    3. Csanádi M,
    4. et al.
    (2019) Benefits of timely care in pancreatic cancer: a systematic review to navigate through the contradictory evidence. Cancer Manag Res 11:9849–9861, doi:10.2147/CMAR.S221427, pmid:31819622.
    OpenUrlCrossRefPubMed
  3. 3.↵
    1. Miles A,
    2. McClements PL,
    3. Steele RJC,
    4. et al.
    (2017) Perceived diagnostic delay and cancer-related distress: a cross-sectional study of patients with colorectal cancer. Psychooncology 26(1):29–36, doi:10.1002/pon.4093, pmid:26868950.
    OpenUrlCrossRefPubMed
  4. 4.↵
    1. Lacey K,
    2. Bishop JF,
    3. Cross HL,
    4. et al.
    (2016) Presentations to general practice before a cancer diagnosis in Victoria: a cross-sectional survey. Med J Aust 205(2):66–71, doi:10.5694/mja15.01169, pmid:27456447.
    OpenUrlCrossRefPubMed
  5. 5.↵
    1. Lyratzopoulos G,
    2. Neal RD,
    3. Barbiere JM,
    4. et al.
    (2012) Variation in number of general practitioner consultations before hospital referral for cancer: findings from the 2010 National Cancer Patient Experience Survey in England. Lancet Oncol 13(4):353–365, doi:10.1016/S1470-2045(12)70041-4, pmid:22365494.
    OpenUrlCrossRefPubMed
  6. 6.↵
    1. Walter FM,
    2. Mills K,
    3. Mendonça SC,
    4. et al.
    (2016) Symptoms and patient factors associated with diagnostic intervals for pancreatic cancer (SYMPTOM pancreatic study): a prospective cohort study. Lancet Gastroenterol Hepatol 1(4):298–306, doi:10.1016/S2468-1253(16)30079-6, pmid:28404200.
    OpenUrlCrossRefPubMed
  7. 7.↵
    1. Stapley S,
    2. Peters TJ,
    3. Neal RD,
    4. et al.
    (2012) The risk of pancreatic cancer in symptomatic patients in primary care: a large case-control study using electronic records. Br J Cancer 106(12):1940–1944, doi:10.1038/bjc.2012.190, pmid:22617126.
    OpenUrlCrossRefPubMed
  8. 8.↵
    The QCancer Tool. accessed. https://qcancer.org/. Oct 2023.
  9. 9.↵
    1. Thompson B,
    2. Philcox S,
    3. Devereaux B,
    4. et al.
    (2021) A decision support tool for the detection of pancreatic cancer in general practice: a modified Delphi consensus. Pancreatology 21(8):1476–1481, doi:10.1016/j.pan.2021.08.007, pmid:34483054.
    OpenUrlCrossRefPubMed
  10. 10.↵
    1. National Institute for Health and Care Excellence (NICE)
    (2023) Suspected cancer: recognition and referral [NICE guideline NG12]. accessed. https://www.nice.org.uk/guidance/ng12/chapter/Recommendations-organised-by-site-of-cancer#upper-gastrointestinal-tract-cancers. 2 Feb 2025.
  11. 11.↵
    1. Hippisley-Cox J,
    2. Coupland C
    (2012) Identifying patients with suspected pancreatic cancer in primary care: derivation and validation of an algorithm. Br J Gen Pract 62(594):e38–e45, doi:10.3399/bjgp12X616355, pmid:22520674.
    OpenUrlAbstract/FREE Full Text
  12. 12.↵
    1. Burmeister EA,
    2. OʼConnell DL,
    3. Beesley VL,
    4. et al.
    (2015) Describing patterns of care in pancreatic cancer: a population-based study. Pancreas 44(8):1259–1265, doi:10.1097/MPA.0000000000000384, pmid:26262591.
    OpenUrlCrossRefPubMed
  13. 13.↵
    1. Chiang PP-C,
    2. Glance D,
    3. Walker J,
    4. et al.
    (2015) Implementing a QCancer risk tool into general practice consultations: an exploratory study using simulated consultations with Australian general practitioners. Br J Cancer 112(Suppl 1):S77–S83, doi:10.1038/bjc.2015.46, pmid:25734392.
    OpenUrlCrossRefPubMed
  14. 14.↵
    1. Nicholson BD,
    2. Hamilton W,
    3. O’Sullivan J,
    4. et al.
    (2018) Weight loss as a predictor of cancer in primary care: a systematic review and meta-analysis. Br J Gen Pract 68(670):e311–e322, doi:10.3399/bjgp18X695801, pmid:29632004.
    OpenUrlAbstract/FREE Full Text
  15. 15.↵
    1. Wiering B,
    2. Lyratzopoulos G,
    3. Hamilton W,
    4. et al.
    (2022) Concordance with urgent referral guidelines in patients presenting with any of six “alarm” features of possible cancer: a retrospective cohort study using linked primary care records. BMJ Qual Saf 31(8):579–589, doi:10.1136/bmjqs-2021-013425, pmid:34607914.
    OpenUrlAbstract/FREE Full Text
  16. 16.↵
    1. Rao G,
    2. Ufholz K,
    3. Saroufim P,
    4. et al.
    (2023) Recognition, diagnostic practices, and cancer outcomes among patients with unintentional weight loss (UWL) in primary care. Diagnosis (Berl) 10(3):267–274, doi:10.1515/dx-2023-0002, pmid:37080911.
    OpenUrlCrossRefPubMed
  17. 17.↵
    1. Nicholson BD,
    2. Aveyard P,
    3. Bankhead CR,
    4. et al.
    (2019) Determinants and extent of weight recording in UK primary care: an analysis of 5 million adults’ electronic health records from 2000 to 2017. BMC Med 17(1), doi:10.1186/s12916-019-1446-y, pmid:31783757. 222.
    OpenUrlCrossRefPubMed
  18. 18.↵
    1. Ali S,
    2. Na R,
    3. Tuesley K,
    4. et al.
    (2022) The association between diabetes mellitus of different durations and risk of pancreatic cancer: an Australian national data-linkage study in women. Cancer Epidemiol 81:102266, doi:10.1016/j.canep.2022.102266, pmid:36240705.
    OpenUrlCrossRefPubMed
  19. 19.↵
    1. Ali S,
    2. Coory M,
    3. Donovan P,
    4. et al.
    (2024) Association between unstable diabetes mellitus and risk of pancreatic cancer. Pancreatology 24(1):66–72, doi:10.1016/j.pan.2023.11.009, pmid:38000983.
    OpenUrlCrossRefPubMed
  20. 20.↵
    1. Keum N,
    2. Ha KH,
    3. Bao Y,
    4. et al.
    (2018) Long-term patterns of fasting blood glucose levels and pancreatic cancer incidence. Cancer Causes Control 29(1):135–142, doi:10.1007/s10552-017-0988-6, pmid:29197995.
    OpenUrlCrossRefPubMed
  21. 21.↵
    1. Mueller AM,
    2. Meier CR,
    3. Jick SS,
    4. Schneider C
    (2022) Characterization of the deterioration of diabetes control in patients with a subsequent diagnosis of pancreatic cancer: a descriptive study. Pancreatology 22(3):387–395, doi:10.1016/j.pan.2022.03.012, pmid:35314354.
    OpenUrlCrossRefPubMed
  22. 22.↵
    1. Martinez-Gutierrez J,
    2. Somsasundaram K,
    3. Bernardes CM,
    4. et al.
    (2026) Clinical decision support tool for early pancreatic cancer detection in primary care: simulation study. JMIR Form Res 10, doi:10.2196/79209. e79209.
    OpenUrlCrossRef
  23. 23.↵
    1. Chima S,
    2. Reece JC,
    3. Milley K,
    4. et al.
    (2019) Decision support tools to improve cancer diagnostic decision making in primary care: a systematic review. Br J Gen Pract 69(689):e809–e818, doi:10.3399/bjgp19X706745, pmid:31740460.
    OpenUrlAbstract/FREE Full Text
Back to top

Latest Articles

Download PDF
Download PowerPoint
Email Article
Citation Tools
The sensitivity of decision support tools for identifying patients with pancreatic cancer: an observational study
Rachel E Neale, Susan J Jordan, Bridie Thompson, Christina M Bernardes, Judi Adams, Christopher Baggoley, Savio George Barreto, Catherine M Baxter, Daniel Croagh, Benedict Devereaux, Jon Emery, Louisa G Collins, Rajit Gilhotra, Paul Grogan, Luke Hourigan, Javiera Martinez-Gutierrez, Andrew J Metz, Stephen Philcox, Meena Rafiq, Joel Rhee, Silja Schrader, Michelle Stewart, John Windsor, John Zalcberg, Mary Waterhouse
BJGP Open 30 June 2026; BJGPO.2025.0142. DOI: 10.3399/BJGPO.2025.0142
del.icio.us logo Facebook logo Mendeley logo Bluesky logo

Jump to section

Keywords

Intended for Healthcare Professionals
 
 

British Journal of General Practice