Interventions to improve timely cancer diagnosis: an integrative review

Abstract

Cancer will affect more than one in three U.S. residents in their lifetime, and although the diagnosis will be made efficiently in most of these cases, roughly one in five patients will experience a delayed or missed diagnosis. In this integrative review, we focus on missed opportunities in the diagnosis of breast, lung, and colorectal cancer in the ambulatory care environment. From a review of 493 publications, we summarize the current evidence regarding the contributing factors to missed or delayed cancer diagnosis in ambulatory care, as well as evidence to support possible strategies for intervention. Cancer diagnoses are made after follow-up of a positive screening test or an incidental finding, or most commonly, by following up and clarifying non-specific initial presentations to primary care. Breakdowns and delays are unacceptably common in each of these pathways, representing failures to follow-up on abnormal test results, incidental findings, non-specific symptoms, or consults. Interventions aimed at ‘closing the loop’ represent an opportunity to improve the timeliness of cancer diagnosis and reduce the harm from diagnostic errors. Improving patient engagement, using ‘safety netting,’ and taking advantage of the functionality offered through health information technology are all viable options to address these problems.

Introduction

Cancer is a leading cause of death, and nearly 40 % of United States residents will ultimately be diagnosed with cancer in their lifetime [1], 2]. Although the great majority of cancers are diagnosed promptly and accurately, missed and delayed diagnoses of cancer are estimated to affect roughly one in every five patients [3]. These delays compound the suffering of patients who receive an unwelcome diagnosis and, in many cases, contribute to potentially avoidable harm due to delayed treatment. In many cases of delayed cancer diagnosis, disease progression may actually preclude use of the most effective treatment regimens.

Most patients with cancer will be cared for in ambulatory settings, especially primary care, before the diagnosis is made. New and improved diagnostic tests will likely improve the earlier diagnosis of cancer in the future, but at the present time, diagnostic delays can be conceptualized as a breakdown at one or more points in the cancer diagnostic process as shown in Figure 1. The model is derived from prior models outlining general and cancer-specific diagnostic pathways [4], [5], [6], [7], [8]. Patients usually enter the cancer diagnostic pathway in one of three ways: (1) from an incidental finding on a lab or medical imaging result, (2) through cancer screening programs (although even in cancers for which screening strategies exist, these do not account for the majority of new diagnoses [9]), or, most often (3), upon investigation of new symptoms or signs, such as blood in the stool, a breast mass, or hemoptysis [10], 11].

Figure 1:

The cancer diagnostic pathway: a model to understand the diagnostic process in cancer.

These various points of entry into the diagnostic pathway create different challenges for clinicians who see these patients in the ambulatory setting. What is common to all three presentation modes is the need to ensure that follow-up tests and consults are completed, and that these results are reviewed and acted upon in a timely manner. “Closing the loop” has been used to describe procedures to ensure follow-up of diagnosed test results [12] and has been elaborated more recently to include other types of follow-up (such as clinical follow-up of suspicious symptoms [13]). The patient’s clinical team must ensure that these loops are closed in a timely manner, and patients with access to their medical records can play an important assistive role as well [14].

A wide range of factors contribute to delayed and missed cancer diagnosis with an equally wide range of proposed solutions [15]. The goal of this integrative review [16], 17] is to summarize the current evidence regarding the frequency of and contributing factors to missed and delayed diagnosis, as well as to present evidence to support possible prevention strategies. We focus in particular on three cancer types (breast, lung, and colorectal cancer) as these all have high incidence [18] and are believed to have appreciable rates of diagnostic error [19].

Methods

We looked for evidence pertaining to missed and delayed diagnosis of breast, lung, and colorectal cancers using a structured search of English language publications in PubMed, Embase, and CINAHL for the period of 2014–2022. We also sought evidence relevant to the primary care setting where many of these missed opportunities in diagnosis occur, although we did not exclude articles from other relevant ambulatory settings. We included search terms related to telemedicine and telediagnosis because of the growth of telemedicine in primary care over the past several years.

In addition to the database search process described above, four researchers (AB, BW, MG, RM) manually searched additional sources from the grey literature, including PSNet (Agency for Healthcare Research and Quality), websites of professional organizations (e.g., Institute for Healthcare Improvement, American Telemedicine Association, Society to Improve Diagnosis in Medicine), and general web searches using the terms “telemedicine,” “safety,” and “diagnosis.” Relevant publications were also identified from the bibliographies of Improving Diagnosis in Health Care, from Recognizing Excellence in Diagnosis: Recommended Practices for Hospitals [20], and from the personal collection of over 3,000 publications on diagnostic error collected by one of the authors (MG). Finally, the research team perused the references cited in full-text articles retrieved for additional relevant publications (snowball sampling).

Results

We identified 493 relevant publications from the combined search strategies described above. Results are summarized thematically by prevalence and risk factors for diagnostic error and potential interventions to improve diagnosis, and by cancer type (cancer in general and breast, lung, and colorectal cancers specifically). We identified the concept of ‘closing the loop’ as a cross-cutting theme.

Delays in cancer diagnosis

Delays in cancer diagnosis are common [21], 22]. Studies from the UK indicate that the median interval from presentation to diagnosis is 40 days, but significant delays were reported in 22–24 % of patient cases [21], 23]. A systematic review summarized the impact of diagnostic delay for seven major cancer types and concluded that for every month the diagnosis is delayed, mortality increases 6–8 % [24]. Additionally, missed and delayed cancer diagnoses are the leading causes for outpatient malpractice claims in the United States [25].

Breast cancer

A large, single-institution retrospective cohort study found that delayed diagnosis was more common when women self-identified an abnormality such as a palpable breast mass or other symptoms, compared with being identified by an abnormal screening mammogram [26]. In a cohort study of 585 U.S. women with breast cancer under the age of 40, a population generally ineligible for routine screening, 17 % reported delayed care seeking and 12 % reported care delays [27].

The World Health Organization’s Breast Health Global Initiative target for disease confirmation after initial presentation is 60 days [28]. This target is often missed; a systematic review found that lack of follow-up after an abnormal mammogram within the next 6 months occurs in 27–72 % of cases [29], 30]. Women with lower educational attainment and women from minority racial and ethnic groups were found to be disproportionately affected by delayed follow-up of abnormal screening mammograms [30]. In a study looking at contributing factors, the most common reason was that follow-up was ordered but never scheduled [31]. Less than 10 % of delays were attributed to patients declining follow-up.

Lung cancer

In a survey of 135 specialists in Australia, 80 % of respondents thought a diagnosis of lung cancer should be obtained within 14 days of first clinical suspicion, but only 56 % estimated that this occurred, mainly due to delays in the primary care setting [32]. Another study suggested that delays in diagnosis of lung cancer in primary care are due to factors preventing expedient referral to specialists [33]. A retrospective cohort case review study conducted at two tertiary Veteran Affairs medical centers identified missed opportunities in 222 out of 587 (37.8 %) patients with lung cancer. The median time to diagnosis rose from 19 days in cases without missed opportunities to 132 days for cases where missed opportunities had occurred [34].

Colorectal cancer

A retrospective cohort study of newly diagnosed cases of colorectal cancer found that 161 of 513 (31.4 %) had experienced at least one diagnostic missed opportunity [35]. Common contributing factors included deficiencies in the history and physical examination and breakdowns in documenting or following up on red flag signs and symptoms [36], 37]. In one study of patients presenting with advanced colon cancer, 18 % had a history of rectal bleeding that was not promptly investigated [38].

Opportunities to improve timely cancer diagnosis

Delays that occur after presentation are prime targets for improving the timeliness of cancer diagnosis. We identified four areas, all under the purview of ambulatory practice, where improvement opportunities have been identified:

Closing the loop on uncertain presentations and non-specific symptoms

Many patients with cancer first present to primary care providers with non-specific symptoms [39]. Compared to age- and sex-matched controls, patients with cancer pursue general practitioner consultations twice as often during the three months prior to their cancer diagnosis [40]. Studies have found that roughly one in five patients with cancer consulted with their general practitioner three or more times for symptoms that were related to the cancer before being referred to a specialist for diagnostic evaluation [41], 42], and an increased risk of cancer diagnosis is found in multiple studies of suddenly increased healthcare utilization [43]. Even when cancer is considered as a possibility, delays in evaluation are common. Hardy et al. concluded that roughly half of patients with delayed cancer diagnosis experienced delays in receiving the appropriate tests or consults [4].

Additional evidence pointing to missed opportunities for diagnosis in primary care emerges from studies of emergency presentations of cancer [44], 45]. Roughly one-third of cancer patients may have an emergency presentation, ranging from 12 % of patients with colorectal cancer to 46 % of patients with pancreatic cancer [45]. These patients are older, have more advanced stages of cancer at presentation, and have poorer survival. In a study of patients with emergency presentations of cancer, two-thirds had recent ambulatory care visits, and one-third had 3 or more such visits [44].

Studies of missed opportunities for diagnosis in primary care have identified a wide range of factors that contribute to delays in the primary care setting [46]. These can be mapped out using a systems-engineering approach (Figure 1), emphasizing the many possible breakdown points in the cancer diagnostic pathway that involve primary care [13]. Delays begin when symptomatic patients delay seeking care or experience delays in engaging the healthcare system.

Ineffective communication during the clinical encounter can also lead to delayed diagnosis. Almost one-third of videotaped analyses of physicians obtaining the patient’s medical history identified misalignment between what the patient was trying to convey about their symptoms and the physician’s interpretation of what was said [47]. Clinical judgment regarding non-specific symptoms that were possibly suggestive of cancer, and ‘gut feelings’ of possible cancer, were identified as key factors in several studies. Having a ‘gut feeling’ of cancer doubled the odds of a cancer diagnosis within 6 months [4]. Complicating the management of patients with nonspecific symptoms is the reality that the vast majority of these patients do not have cancer. Even patients with ‘red flag’ presentations such as rectal bleeding have less than a 10 % likelihood of harboring a related malignancy [48], creating challenges in deciding who needs urgent and comprehensive evaluation and who does not.

Besides emphasizing the importance of improving the initial patient encounter and the need to follow these patients at appropriate intervals, system-based interventions, such as the use of urgent-referral pathways, may also be effective [49], 50]. Okoli et al. reviewed ten types of interventions to improve early cancer diagnosis of symptomatic patients and identified several effective practices, including the use of rapid referral pathways, standardized care pathways, nurse navigators, and technology-based approaches to monitor follow-up [49]. Another effective practice to manage uncertainty is “safety-netting” [51], a management approach that focuses on establishing clear communication during the patient encounter, agreeing on a plan for follow-up, and outlining actions to be taken by both the patient and the clinician to ensure follow-up takes place.

Ultimately, better tools and tests are needed to identify which patients with non-specific presentations are at highest risk for cancer, which will require large epidemiologic studies looking at the positive predictive value of various presentations [52]. As an example, adult patients seen in primary care settings with non-specific abdominal pain or bloating have an overall risk of developing cancer within 1 year of roughly 2 %, but this risk increased substantially with abnormal lab tests: 10 % with raised ferritin, 9 % with low albumin, 8 % with raised platelets, 6 % with raised inflammatory markers, and 4 % with anemia [53].

Closing the loop on incidental findings

Incidental findings on imaging present the opportunity to diagnose and treat malignancies at the earliest possible stage. One in every twenty-five lung cancers is first detected as an incidental finding [54]. Unfortunately, this opportunity is often missed due to delays and failures in appropriate follow-up [55]. For example, in a large study of almost 3 million imaging examinations, 10.6 % of the imaging interpretations contained a recommendation for follow-up imaging, but adherence was only 58 % within 14 months [56]. Breakdowns in follow-up are most common for imaging performed in emergency department settings for children and the elderly [57].

Roughly 4 % of cancers are diagnosed from incidental findings on radiologic imaging [54]. Incidental findings are encountered in approximately one in three CT [58] and MRI [59] studies. Roughly two-thirds of incidental imaging findings are considered actionable [55], and most of these turn out to be false positives or benign conditions. Only one in twenty incidental findings are ultimately diagnosed as a malignancy [60], 61].

Improving communication amongst radiologists, front-line clinicians, and patients is the focus of most recommendations to improve the management of incidental findings [62], and a range of health information technology (IT)-based applications to track and ensure follow-up actions are emerging. It is not sufficient to simply include follow-up recommendations in a report: “a mandatory phone call from the radiologist to the ordering physician should be instituted for all important or potentially actionable unanticipated findings” [63]. The federal Mammography Quality Standards Act codified the requirement for radiologists to communicate suspicious findings directly to patients and serves as a role model for how direct communication could similarly improve the follow-up of incidental findings. The “Failsafe” program at Hershey Medical Center uses a telemedicine approach to complement these practices. The program leverages health IT applications that identify patients with important incidental findings, facilitate a phone call to these patients, and enable tracking of the process and ultimate completion of the follow-up [64].

Interventions that target incidental findings include recommendations to standardize the nomenclature and classification of incidental findings and the language used to convey the importance of a given finding [60], 63], 65]. For example, new fields in the radiology report could include: (1) What is recommended? (e.g., repeat a test or perform a different test), and (2) What is the time frame in which to perform the recommended action? After one institution implemented this approach, 87 % of patients with incidental findings complied with recommended follow-up [66]. A similar intervention focused on patients imaged as part of an emergency visit, 95 % of patients received appropriate follow-up communication and had documented follow-up plans [67].

The argument has also been made that the pursuit of incidental findings on medical imaging examinations is costly, time-consuming, anxiety-producing, and potentially dangerous in terms of additional exposure to ionizing radiation and possibly invasive procedures [68], 69], especially in relation to the relatively low rate with which incidental findings ultimately turn out to be malignancies [47], 48]. The pursuit of incidental findings also engenders overdiagnosis of small and slow-growing cancers that have low potential for harm. Carefully considered guidelines from the American College of Radiology provide recommendations on when and how to pursue incidental findings that balance these benefits and hazards [70].

Closing the loop on testing and positive screening tests

Cancer diagnosis very often hinges on the results of diagnostic tests. Improving the timeliness and reliability of acknowledging and acting on these results would have a large impact on improving cancer diagnosis. The frequency of missed test results varies considerably from study to study, ranging from 6 to 62 % [71], 72]. Tests requiring “send outs” to specialized labs and tests pending at discharge from a hospitalization have especially high rates of communication failure [73].

Similar problems are encountered regarding appropriate follow-up after imaging as noted previously [56]. Minimizing the use of vague or conditional language in imaging reports may facilitate follow-up of abnormal imaging results [66]. Recent publications have highlighted the importance of closing the loop on suspicious imaging results [62], 66], 74].

Ensuring appropriate review and follow-up of tests is a significant problem. A survey of 275 primary care providers found widespread dissatisfaction with their existing processes for managing abnormal test results, citing patients’ practical and logistical barriers to pursuing follow-up and a paucity of tools within their electronic record systems for managing these situations [75]. A proposal for clinicians to audit their own practices for test result follow-up showed a small improvement in rates of appropriate follow-up [76]. However, these problems reflect overarching limitations such as poor communication and systems failures, which involve all of the parties involved in the diagnostic process as well as the organization’s electronic health record system.

Many breakdowns in test result communication reflect ambiguity in defining responsibility for follow-up, absence of backup or fail-safe processes, and the unfortunate assumption on the part of the patient that “no news is good news” [77]. Several comprehensive reviews have summarized recommendations for improving test result follow-up [78], [79], [80]. Both clinician groups [78] and a wide range of organizations have also outlined suggestions and best practices for test results management that address these issues [12], [81], [82], [83], [84]. Among these is the recommendation that organizations adopt comprehensive and clear communication policies that unambiguously define who is responsible for test result follow-up [85]. Ideally, these policies would adopt a ‘chain of custody’ mindset, similar to that in many transaction-based industries, such as shipping companies or airlines [86], 87]. In this approach, diagnostic test results are electronically tracked to ensure timely delivery. These solutions could include additional features like alerts, notes, and patient instructions [88].

Progress in adopting and using these recommendations effectively has been mixed. One systematic review found that many EHR-based notifications and related health IT innovations had little impact on time to further action and diagnostic resolution [89]. Moreover, these interventions come at the cost of increased workload [90], [91], [92]. Qualitative observations of implementing test result management systems underscore the range of factors affecting implementation, the potentially pivotal role of less frequently acknowledged nonclinical team members, and contextual factors that determine the gap between ideal and actual use.

One promising development in the detection of potential diagnostic delays is the development of algorithms that identify patients at risk for lack of follow-up for abnormal test results [79], 93]. These algorithms identify patients with abnormal findings who have not received timely follow-up within a set time period. This approach is used by the SureNet system at Kaiser Permanente health system [94]. Another approach, described as the “Backstop” system, is a similar tracking system for imaging. One month after the recommended follow-up test is past due, the original radiology report is re-sent to the primary care provider if expected follow-up action could not be detected. After two months, a navigator telephones the primary care provider’s practice. After three months, the radiologist telephones the primary care provider directly. Use of these interventions increased the proportion of completed follow-up imaging studies from 43 to 54 % [95].

Closing the loop on consults and referrals

Vulnerabilities related to consults and referral processes in regard to cancer diagnosis have been documented and discussed in several studies [96], [97], [98]. In one study involving 100,000 consults and referrals across 34 primary care clinics, only about one-third were associated with a subsequent documented and completed specialty visit [99]. Factors associated with failure included ambiguous documentation, inability to contact patients for scheduling, patient refusal of specialty care, geographical barriers, and out-of-network referrals.

The Institute for Healthcare Improvement has published a comprehensive toolkit to improve management of consults and referrals [100], and recommendations are also available from others [101], 102]. In a successful field trial in another specialty (cardiology), the research team engaged primary care providers and specialists to develop and implement a change package for referral management. From pre- to post-intervention, a higher proportion of referrals was closed within seven days of the specialist visit (from 40 to 70 %), and a higher proportion of specialist reports were found to answer the referral question (from 50 to 75 %) [103]. Other outcomes included improved provider satisfaction and perceived improvements in the communication among both providers and patientst.

Patient engagement to close the loop

To complement the efforts of organizations to successfully follow-up on abnormal tests and imaging studies, there is great potential for patients to participate in this effort, promoting safety in their own care [64]. Not all patients, however, may be willing or able to take appropriate next steps, and the act of notifying a patient about an abnormal finding does not guarantee that the information will be understood, acted upon, or even recalled [64], 71]. A study of 1,152 patients in Singapore with positive screening tests for colon cancer found that only 56 % agreed to a follow-up consultation with a gastroenterologist [104]. About 40 % of the patients who initially declined follow-up went on to arrange follow-up on their own. Of those who did not, the most frequently reported barriers to follow-up included denial of a significant result (41.8 %), knowledge deficits about the meaning and significance of the abnormal test result (34.5 %), unwillingness to see an unfamiliar provider (20 %), and lack of follow-up reminders (20 %). Following a scripted telephone call informed by a health behavior change theory, 20 % of patients lacking follow-up agreed to a gastroenterology consultation.

Discussion

Primary care providers play a critical role in recognizing cancer at the earliest possible stages and guiding patients through the many steps between suspicion and a confirmed diagnosis. For most patients this process works well, and the diagnosis is made quickly. Unfortunately, there are myriad opportunities for the diagnostic process to be delayed or not completed at all, and each of these represents a possibly unacceptable outcome. Our review identified ‘closing the loop’ as a cross-cutting theme underlying the following practices to improve timely diagnosis of cancer:

• –For patients with non-specific symptoms – ensuring follow-up until symptoms resolve or a diagnosis is established;
• –For patients with incidental findings – ensuring these are followed up according to accepted guidelines;
• –For patients with abnormal screening tests – ensuring that appropriate confirmatory tests are completed promptly;
• –For patients with abnormal test results or tests pending at discharge – ensuring that these results are reviewed, communicated, and acted on;
• –For patients for whom specialty consultation is needed – ensuring that the consult or referral is completed, and the impressions are communicated.

Several of the interventions reviewed here are cross-cutting and represent applications of the ambulatory ‘safety net’ concept to catch cases headed for a diagnostic delay [105]. These include improving patient engagement, using patient navigators to monitor and guide care, and employing electronic monitoring systems to detect instances of delay or failure of follow-up. No one solution will completely solve the problem of delayed cancer diagnosis, and each practice will need to consider and prioritize their own interventions. Public health education on cancer risk would be helpful as a general measure, along with more specific campaigns targeting low-income and other marginalized patient populations. Delays in diagnosis often reflect some combination of fear, lack of awareness, and practical barriers (no health insurance, no transportation, etc.) that result in patients not being screened or following up afterwards.

Progress will require addressing the current implementation gap for adopting known practices in ambulatory care, and new incentives and programs are needed [106]. Practices often do not have adequate data to show them how well they are closing the loop. Resources and infrastructure for safety are less developed for ambulatory vs. inpatient settings. Practices will require additional time, incentives, and resources to implement interventions to close the loop. Local champions are needed with dedicated time to identify barriers and opportunities for improvement, and multi-site collaboratives could host research projects to study and compare different solutions. Whitfield et al. have conceptualized the path ahead as addressing five key questions [107]: (Table 1), and seeking answers will require a range of different research approaches, including epidemiologic studies based on medical records data, clinical audit studies of individual patients, and patient surveys.

Table 1:

Questions to direct future research on improving timely cancer diagnosis.

1.	Do healthcare usage patterns suggest earlier diagnosis could be possible? (Studies looking at the pre-diagnostic ‘window’ period)
2.	How does the diagnostic process begin? (Studies looking at prodromal presentations)
3.	How do patients progress from presentation to diagnosis? (Studies looking at progression pathways)
4.	How long does the diagnostic process take? (Studies looking at the various intervals)
5.	Could anything have been done differently to reach the correct diagnosis? (Studies looking at missed opportunities)

In sum, we find that breakdowns and delays are unacceptably common in cancer diagnosis pathway, representing failures to follow-up on abnormal test results, incidental findings, non-specific symptoms, or consults. Interventions aimed at ‘closing the loop’ represent an opportunity to improve the timeliness of cancer diagnosis and reduce the harm from diagnostic errors. Closing these loops represents an actionable approach to improving the timeliness of cancer diagnosis and reducing the harm from diagnostic errors in the United States and internationally.