1. Chalmers I, Bracken MB, Djulbegovic B, Garattini S, Grant J, Gülmezoglu AM, et al. How to increase value and reduce waste when research priorities are set. Lancet. 2014;383(9912):156–65.[PubMed]
2. Entwistle V, Calnan M, Dieppe P. Consumer involvement in setting the health services research agenda: persistent questions of value. J Health Serv Res Policy. 2008;13(suppl 3):76–81.[PubMed]
3. Banfield MA, Barney LJ, Griffiths KM, Christensen HM. Australian mental health consumers’ priorities for research: qualitative findings from the SCOPE for Research project. Health Expect. 2014;17(3):365–75. [PMC free article][PubMed]
4. Staley K. Exploring impact: public involvement in NHS, public health and social care research. INVOLVE, Eastleigh. 2009
5. Brett J, Staniszewska S, Mockford C, Herron-Marx S, Hughes J, Tysall C et al. Mapping the impact of patient and public involvement on health and social care research: a systematic review. Health Expect. 2014;17(5):637–50. [PMC free article][PubMed]
6. Staley K, Hanley B. Scoping research priority setting (and the presence of PPI in priority setting) with UK clinical research organisations and funders. James Lind Alliance. December 2008.
7. Abma TA. Patients as partners in a health research agenda setting the feasibility of a participatory methodology. Eval Health Prof. 2006;29(4):424–39.[PubMed]
8. Caron-Flinterman JF, Broerse JEW, Bunders JFG. The experiential knowledge of patients: a new resource for biomedical research? Soc Sci Med. 2005;60(11):2575–84.[PubMed]
9. Whear R, Thompson‐Coon J, Boddy K, Papworth H, Frier J, Stein K. Establishing local priorities for a health research agenda. Health Expect. 2012 [PMC free article][PubMed]
10. Cheyne H, McCourt C, Semple K. Mother knows best: developing a consumer led, evidence informed, research agenda for maternity care. Midwifery. 2013;29(6):705–12.[PubMed]
11. Fleurence R, Selby JV, Odom-Walker K, Hunt G, Meltzer D, Slutsky JR, et al. How the patient-centered outcomes research institute is engaging patients and others in shaping its research agenda. Health Aff. 2013;32(2):393–400.[PubMed]
12. Hinckley J, Boyle E, Lombard D, Bartels-Tobin L. Towards a consumer-informed research agenda for aphasia: preliminary work. Disabil Rehab. 2014;36(12):1042–50. [PubMed]
13. Schipper K, Dauwerse L, Hendrikx A, Leedekerken J, Abma T. Living with Parkinson’s disease: priorities for research suggested by patients. Parkinsonism Related Disor. 2014;20(8):862–6.[PubMed]
14. Stewart RJ, Caird J, Oliver K, Oliver S. Patients’ and clinicians’ research priorities. Health Expect. 2011;14(4):439–48.[PMC free article][PubMed]
15. Staniszewska S, Adebajo A, Barber R, Beresford P, Brady LM, Brett J, et al. Developing the evidence base of patient and public involvement in health and social care research: the case for measuring impact. Int J Consumer Studies. 2011;35(6):628–32.
16. JLA. The James Lind Alliance guidebook. James Lind Alliance, Oxford. 2013. http://www.JLAguidebook.org. Accessed January 2015.
17. Chalmers I, Atkinson P, Fenton M, Firkins L, Crowe S, Cowan K. Tackling treatment uncertainties together: the evolution of the James Lind Initiative, 2003–2013. J Royal Soc Med. 2013;106(12):0141076813493063.[PMC free article][PubMed]
18. Gadsby R, Snow R, Daly A, Crowe S, Matyka K, Hall B et al. Setting research priorities for type 1 diabetes. Diabetic Med. 2012;29(10):1321–6. [PubMed]
19. Cresswell J, Plano C. Designing and conducting mixed methods research. 2011.
20. Bradley EH, Curry LA, Devers KJ. Qualitative data analysis for health services research: developing taxonomy, themes, and theory. Health Serv Res. 2007;42(4):1758–72.[PMC free article][PubMed]
21. Oliver S, Clarke-Jones L, Rees R, Milne R, Buchanan P, Gabbay J, et al. Involving consumers in research and development agenda setting for the NHS: developing an evidence-based approach. Health Technol Assess. 2004;8(15):154pp.[PubMed]
22. Batchelor J, Ridd M, Clarke T, Ahmed A, Cox M, Crowe S, et al. The Eczema Priority Setting Partnership: a collaboration between patients, carers, clinicians and researchers to identify and prioritize important research questions for the treatment of eczema. British J Dermatology. 2013;168(3):577–82.[PubMed]
23. Elberse J, Laan D, de Cock BT, Teunissen T, Broerse J, de Boer W. Patient involvement in agenda setting for respiratory research in the Netherlands. Eur Respiratory J. 2012;40(2):508–10.[PubMed]
24. Lophatananon A, Tyndale-Biscoe S, Malcolm E, Rippon HJ, Holmes K, Firkins LA, et al. The James Lind Alliance approach to priority setting for prostate cancer research: an integrative methodology based on patient and clinician participation. BJU Int. 2011;108(7):1040–3.[PubMed]
25. Vortruba M, Group S-PS The UK Sight Loss and Vision Priority Setting Partnership (SLV-PSP): vision research questions prioritised by patients and health care professionals. Acta Ophthalmologica. 2013;91(s252):0.
Missed opportunities for diagnosing cancer sooner may occur anywhere in the diagnostic process. On the basis of evidence from retrospective case reviews of cohorts of cancer patients, missed opportunities typically occur in three main phases:
Initial diagnostic assessment (during the clinical encounter between a patient and a doctor, typically, but not exclusively, a generalist). This phase involves history taking, clinical examination and diagnostic reasoning, potentially also leading to specialist referral, test ordering or expectant (‘safety netting’/‘wait and see’) management decisions, or their combination.
Diagnostic test performance and interpretation. This phase involves the process of performing appropriate diagnostic tests (e.g., blood tests, imaging or endoscopy, often at different times and locations) and their appropriate interpretation and associated actions.
Diagnostic follow-up and coordination. This phase includes many activities and tasks required to ‘close the loop’ on test results and referrals made on initial diagnostic assessment.
Patient, provider and system factors can all contribute to the generation of missed opportunities during one or more of the above phases, and missed opportunities in diagnosis often involve more than one contributory factor (Singh et al, 2013a). Complex interactions exist between these factors; for example, both patient and doctor factors could be influenced by system factors (Andersen et al, 2014). Understanding the complex interplay between these factors is important for reducing missed opportunities, thus underscoring the importance of using multi-disciplinary approaches in this area, including perspectives from psychology, human factors (the scientific field that focusses on how people interact with products, tools, procedures and processes) and informatics.
The concept of missed diagnostic opportunities builds on previous theoretical models from psychology. For example, the ‘model of pathways to treatment’ provides a holistic consideration of the journey from symptom onset to diagnosis and treatment initiation, encompassing four distinct intervals (symptom appraisal, help-seeking, diagnostic and pre-treatment intervals), with the diagnostic interval being of relevance to missed opportunities after presentation as considered in this paper. (Walter et al, 2012; Scott et al, 2013).
We use the high-level taxonomy of phases described above toillustrate different types of missed opportunities and related contributing factors to inform and motivate further policy initiatives and research.
Missed opportunities and contributing factors during initial diagnostic assessment
Rigid consultation norms
In some countries (including the UK and Denmark) medical consultation norms encourage patients to consult for ‘one problem at a time’ (which may even have to be declared in advance, before consultation), while the duration of primary care appointments is typically as short as 10 min (National Health Service Information Centre, 2007; Andersen et al, 2014; McCartney, 2014). Further, notable proportions of the public in countries with publicly funded health-care systems worry about consulting for symptoms that may ‘waste the doctor’s time’ (Forbes et al, 2013). Beyond increasing the risk of delayed presentation and help-seeking, such attitudes might also decrease patient resolve to use up consultation time for communicating the full breadth and complexity of their symptoms, thereby increasing the risk of missed opportunities (Andersen et al, 2011).
Inadequate history taking and examination
For several reasons, the full spectrum, nature and duration of symptoms may not be elicited during a primary care encounter. Retrospective medical record reviews of patients diagnosed with cancer indicate that insufficient symptom elicitation or recording and ineffective doctor-patient communication may account for many instances of missed opportunities (Singh et al, 2009a; Jensen et al, 2014). Time pressures, either real or perceived, may impede doctors to obtain a thorough history or elicit clinical signs when present (Andersen et al, 2014; McCartney, 2014). Other factors described below might also contribute.
An increasing number of cancer patients in Europe and North America have limited proficiency in the first language of their resident country. In such circumstances, lack of interpretative support may impede effective patient–doctor communication, with some epidemiological evidence suggesting that suspecting the diagnosis of cancer is less prompt (i.e., requiring a greater number of pre-referral consultations) in older ethnic minority patients with symptoms (Lyratzopoulos et al, 2012).
Cognitive factors impeding optimal initial clinical assessment and reasoning
Firmly suspecting the diagnosis of cancer during a single clinical encounter is difficult, as symptoms and signs are rarely pathognomonic and may also be seen early in their development (Jones et al, 2007; Hamilton 2009; Jones et al 2009). Furthermore, as public awareness campaigns, by their very nature, encourage larger proportions of persons with symptoms to consult, the already low positive predictive value of symptomatic presentations in primary care for cancer may decrease further. A range of factors may, however, make diagnostic reasoning even more challenging. These include the following:
Cognitive biases: several types of such biases exist, including anchoring bias (focusing exclusively on a single item of information), availability bias (over-reliance on already known or easily available information) and ‘commitment to a steer’ (i.e., initial diagnostic impressions), which can impede diagnostic reasoning (Kostopoulou et al, 2012; Croskerry 2013). In a study of diagnostic errors in UK primary care, biases at the initial framing of the problem were related to errors at the end of the diagnostic process (Balla et al, 2012). In addition to operating during the clinical encounter, these biases could also lead to the misinterpretation of diagnostic test results (Singh et al, 2012a).
Co-morbidity: consideration of a cancer diagnosis is particularly challenging in the presence of other known non-cancer co-morbid conditions. Many older patients (the age group at higher risk for cancer) are multi-morbid (Barnett et al, 2012). In these patients, symptoms compatible with the known cause of chronic morbidity could be easily thought to reflect the pre-existing disease rather than a new problem (Mitchell et al, 2013).
Unfamiliarity with cancer presentations: patients with a new diagnosis of cancer are infrequent in general practice. For example, in the UK a full-time general practitioner on average might encounter only between 5 and 10 new cases in a year, amid thousands of patients with other conditions. Beyond unfamiliarity, ‘epidemiological optimism’ bias can make prompt suspicion of the diagnosis of cancer even harder in low-risk patient groups even when they complain of symptoms that may be due to cancer. Such groups include young persons and certain socio-demographic groups within specific cancers (e.g., women who present with visible haematuria; Lyratzopoulos et al, 2012, 2013b; Nicholson et al, 2014).
Access and system capacity constraints
These are often expressed as long waiting times. An indirect consequence of prolonged waiting times is that they de facto increase the disease severity threshold for referral decisions—i.e., capacity constraints influence doctor-decision making. This may be a particular challenge for publicly funded systems where demand management functions are implicitly delegated to primary care services (Vedsted and Olesen 2011; Brown et al, 2014). Geographical barriers, such as distance to diagnostic centres, may also be relevant, although evidence on such associations is needed.
The positive predictive value of signs and symptoms for cancer is low; only a few have values >5% in patients presenting in general practice (Shapley et al, 2010). Consequently, most patients investigated for suspected cancer will not have the disease; for example, in the UK among patients who are referred to specialists for suspected cancer, about 90% will be found not to have cancer (Meechan et al, 2012). Likely peer pressure by specialists or hospital managers may increase reluctance by primary care doctors to refer patients if there is intolerance of low ‘diagnostic hit rates’, increasing the risk of missed opportunities, although evidence to further establish the role of such dynamics would be desirable.
Missed opportunities and contributing factors during diagnostic test performance and interpretation
Patient non-adherence with recommended tests and lack of system resilience towards such (‘no show’) events
Patients may not adhere to prescribed investigation plans, either by not attending recommended investigations or by not preparing for them. For example, patient non-adherence with suggested colonoscopy investigation can lead to post-referral diagnostic delays in colorectal cancer (Singh et al, 2012b). In Denmark, among all missed opportunities, 16% were attributable to patients not showing up (Jensen et al, 2014). Different factors may be implicated in ‘no show’ events. Evidence from patients not adhering to screening colonoscopy appointments implicates emotional barriers (such as fear of an adverse diagnosis, or fear of procedure-related pain or complications) and logistical or communication barriers (Denberg et al, 2005), and similar factors may be applicable to diagnostic colonoscopies. When ‘no show’ events do occur, communication with the patient and a review of the patient’s management should be automatically triggered to reschedule investigations or initiate alternative management.
Diagnostic testing process complexity
The diagnosis of cancer typically requires a sequence (‘chain’) of tests and procedures (e.g., blood tests, imaging, endoscopy leading to tissue sampling and pathology reporting). These tests are by necessity often carried out at different locations and times, whereas generalist and specialist doctors often work in different settings, adding levels of complexity to the diagnostic process. This high degree of complexity multiplies the risk of delays and/or erroneous decision making at different steps in the process by a factor proportionate to the number of distinct tests required. Further, the distribution of the diagnostic process (in space and time, and between primary care and specialist care, often including different departments and locations) increases both time lags between different steps in the process and risks of miscommunication (Jensen et al, 2014), untimely communication or lack of follow-up of important test results. This has, for example, in Denmark, led to the formation of diagnostic units that enable the conduct of multiple required tests and specialist assessments ‘within 1 day/under one roof’, and similar services are currently being developed in the UK. Such strategies will need to be evaluated for their effectiveness. Earlier work by the National Patient Safety Agency has recommended that the health service needs to ‘identify, review and disseminate current good practice in the process of ordering, managing and tracking tests and test results’ (NPSA, 2010).
Inadequacies in the investigation strategy (initially negative tests in the presence of ongoing symptoms or diagnostic suspicion)
This may occur when the suspicion of cancer is correctly raised but decisions on planned investigations are sub-optimal or inadequate (Jensen et al, 2014). This scenario may be more likely for cancers sharing many common symptoms (e.g., cancers of pelvic/abdominal organs). For example, a patient with abdominal symptoms is investigated with a colonoscopy that is negative, and this finding is initially interpreted as bringing ‘diagnostic closure’, but the patient has persistent symptoms and is subsequently found to have cancer of another abdominal organ (pancreas, liver or ovary). Another example can be provided by false-negative chest X-ray findings in patients with suspected lung cancer (Bjerager et al, 2006; Stapley et al, 2006). Such circumstances can clearly prolong diagnostic intervals by providing (temporary) false reassurance (Singh et al, 2012a). A much more challenging situation occurs when correct tests have been carried out but the results are falsely interpreted as negative, without adequate fail-safe or back-up re-assessment mechanisms being present (Singh et al, 2012a; Middleton et al, 2014).
Missed opportunities and contributing factors during follow-up and coordination
If appropriately empowered, patients’ active role in the diagnostic process can minimise the risks of missed opportunities. How this potential can be harnessed should be a priority for future research (McDonald et al, 2013). The National Patient Safety Agency report on diagnostic delays in cancer recommended that the health service ‘develop methods for empowering patients on a cancer diagnostic pathway’ (NPSA, 2010). However, currently:
Many patients do not feel empowered to seek out the results of tests performed on them, or do not know how to do so. Patients may also be ‘reassured’ by lack of follow-up by doctors/the health-care system, interpreting lack of communication to mean that ‘all is normal’ in instances when this is not the case. This emphasises the importance of passive (‘open door’) or active (fixed interval, e.g., 3-week clinical review) follow-up as part of safety-netting strategies (Almond et al, 2009).
Patients also might not be willing to re-consult or seek a second medical opinion despite doubting the certainty of their diagnosis and persistent or worsening symptoms (Birt et al, 2014; McDonald et al, 2013).
Another common occurrence is patients not returning for ‘fail-safe/safety-netting’ visits planned as part of expectant management strategies, and/or when they experience persistent, worsening or new symptoms (Singh et al, 2012a; Mitchell et al, 2013). Factors similar to those involved in non-adherence with diagnostic investigations may be implicated (see ‘Missed opportunities and contributing factors during diagnostic test performance and interpretation’, above). Robust mechanisms for identification of such occurrences and contact with patients are required.
Over-reliance on patients to ‘call back’
Doctors often believe that patients will call if they do not feel better or new symptoms develop, and often assume that the diagnosis they had recently given had been correct if they do not hear otherwise (Singh and Sittig, 2014). Ensuring timely patient follow-up could also help prevent missed opportunities that relate to coordination failures between different clinics, hospital departments and general practices (Mitchell et al, 2013). Proactive follow-up systems and protocols that leverage information technology might be needed to minimise the risk of such missed opportunities ‘at the last hurdle’ (Murphy et al, 2014).
Lack of appreciation or follow-up of abnormal test results
Increasingly recognised in ‘electronic health record-enabled’ health-care systems are instances of failure to recognise and act on abnormal tests related to cancer (Murphy et al, 2014). Many reasons could contribute to such occurrences—for example, physician ‘alert fatigue’ or ambiguities about the health-care professional who is ‘in charge’ of the patient and responsible for follow-up (Singh et al, 2009b). Informational continuity and clarity of accountability for the patient as they progress through the diagnostic pathway poses remarkable challenges (Press, 2014). Because modern health care is delivered by teams rather than by individuals, factors relating to team dynamics and ‘distributed cognition’ could also be relevant; these include low staff morale, poor communication between team members and limited ‘situational awareness’ of diagnostic safety (Singh et al, 2012a).