The diagnostic pathway for patients with interstitial lung disease: a mixed-methods study of patients and physicians
•,,,,,,.
...
Abstract
Objectives The diagnostic process for patients with interstitial lung diseases (ILD) remains complex. The aim of this study was to characterise the diagnostic care pathway and identify barriers and potential solutions to access a timely and accurate ILD diagnosis.
Design This mixed-method study was comprised of a quantitative chart review, patient and physician surveys and focus groups.
Results Chart review was completed for 97 patients. Median time from symptom onset to ILD diagnosis was 12.0 (IQR 20.5) months, with 46% diagnosed within 1 year. Time from first computed tomography (CT) scan to respirology referral was 2.4 (IQR 21.2) months. Referrals with a prior CT were triaged sooner than referrals without (1.7±1.6 months vs 3.9±3.3 months, p=0.013, 95% CI 0.48 to 2.94). On patient surveys (n=70), 51% felt that their lung disease was not recognised early enough. Commonly reported challenges to timely diagnosis included delayed presentation to primary care, initial misdiagnoses and long wait-times for specialists. Forty-five per cent of physicians (n=20) identified diagnostic delays, attributed to delayed presentations to primary care (58%), initial misdiagnoses (67%) and delayed chest imaging (75%). Themes from patient and respirologist focus groups included patient-related, healthcare provider-related and system-related factors leading to delays in diagnosis.
Conclusions This mixed-methods study identified patient and system-related factors that contribute to diagnostic delays for patients with ILD, with most delays occurring prior to respirology referral. ILD awareness and education, earlier presentation to primary care, expedited access to chest imaging and earlier referral to respirology may expedite diagnosis.
What is already known on this topic
Patients with interstitial lung disease (ILD) often experience diagnostic delays. A timely diagnosis is critical as currently available pharmacologic therapies can slow the progression of ILD, but cannot reverse established fibrosis.
What this study adds
This mixed-method study identified patient-reported and physician-reported barriers leading to the diagnostic delay of patients with ILD including delayed presentation to primary care, initial misdiagnoses, delayed chest imaging and long wait-times for specialists.
Using survey and focus group data, we provide patient and physician informed suggestions to improve the diagnostic pathway of patients with ILD.
How this study might affect research, practice or policy
This work will help optimise the diagnostic journey of patients, which will require the engagement of multidisciplinary stakeholders aimed at improving the accuracy of and reducing the timing to an ILD diagnosis.
Introduction
Interstitial lung diseases (ILD) represent a heterogeneous group of disorders characterised by lung parenchymal inflammation and/or fibrosis. Timely diagnosis is important to ensure risk factor mitigation, treatment initiation and prognostication.1 This timely diagnosis is critical as currently available pharmacologic therapies can slow the progression of ILD, but cannot reverse established fibrosis.2 3 An accurate diagnosis may also prevent exposure to harmful or ineffective therapies, and increase access to measures that can improve patients’ quality of life such as supplemental oxygen, pulmonary rehabilitation and palliative care.4
Challenges to early diagnosis include the insidious onset and non-specific symptoms of ILD, which patients may attribute to age or deconditioning, leading to delays in seeking medical attention.5 Limited access to chest imaging and the capacity of specialised ILD centres and multidisciplinary discussions (MDD) may also impede a timely diagnosis.2
The objective of this mixed-methods study was to characterise the diagnostic pathway of patients with ILD and identify sources of diagnostic delay, using data from chart reviews, surveys and focus groups. We sought to better understand the challenges faced by patients in receiving a timely diagnosis, and to identify actionable solutions to improve diagnostic efficiency.
Methods
Study participants
Patient participants were adults ≥18 years with a self-reported ILD diagnosed within 3 years, able to provide consent, proficient in English and with internet access. ILD diagnoses included idiopathic pulmonary fibrosis, connective tissue disease association ILD, fibrotic hypersensitivity pneumonitis, occupational ILD and unclassifiable ILD. Patients with sarcoidosis were not included. Physicians with Royal College certified respirology training, currently practicing in Alberta, Canada and who care for ILD patients were eligible. This study was conducted in a Canadian province which provides universal healthcare to an estimated population of 4.5 million and has two ILD centres.
Data collection
Chart review
Respirologists who had a high frequency of ILD referrals were invited via email to participate in a chart review of newly diagnosed ILD patients between 1 January and 31 December 2019, chosen to avoid the 2020 impact of COVID-19. Data extracted included demographics, referral date, patient reported symptoms, investigations (pulmonary function tests (PFT), arterial blood gases, autoimmune serology, chest radiograph (CXR), CT, bronchoscopy and lung biopsy), treatment history, initial referral diagnoses and final ILD diagnoses (online supplemental figure 1).
Separate patient and physician focus groups were conducted. Participants who completed the online survey were voluntarily recruited by email to participate in a 1 hour focus group conducted over ZOOM. Data analysis was based on grounded theory and reported based on the consolidated criteria for reporting qualitative research6 (online supplemental table 1). Discussions were moderated (AGO and CP) in a semistructured, open-ended approach (online supplemental figure 3). Focus groups were transcribed then analysed using QSR International Pty Ltd. NVivo (V.12). Field notes were documented. A third coder (MK) assisted with consensus. The transcripts were independently analysed by two investigators (AGO and CP). Using open coding, transcripts were read line-by-line, descriptive codes applied and themes developed. A code book was developed from the first focus group. Data saturation was achieved when no new codes were found and themes answered the inquiry. This occurred after the first focus group for physicians and second for patients, whereby one additional focus group was conducted per group. The transcripts were searched a second time to refine the relationships between codes and themes. Direct illustrative quotes were extracted.
Patient and public involvement statement
None.
Statistical analysis
Means, SD, medians and IQR were used to describe continuous variables. Categorical data were reported with descriptive frequencies. Student t-test was performed to compare means, p<0.05 was statistically significant. Analyses were conducted using SAS software (V.8.3). All participants provided informed consent.
Results
Chart review
Of the invited respirologists, 12/39 (31%) participated and contributed data from 100 ILD patients; three had incomplete data, thus 97 patients’ charts were included. Mean patient age at diagnosis was 65.6±16.0 years, 56% were male.
Patients had symptoms for a median of 10.5 (IQR 19.0) months prior to their first respirology visit. The most common symptom was dyspnoea (76%) followed by cough (63%) and fatigue (25%) (table 1); 10% of patients were asymptomatic. Prior to respirology referral, 96% of patients had imaging studies and 75% had PFTs ordered, with a median of 2.0 (IQR 3.0) CXRs and 1.0 (IQR 1.0) CTs. Median time from symptom onset to first CT was 5.0 (IQR 12.1) months. Time from first CT to respirology referral was 2.4 (IQR 21.2) months (figure 1A). Fifteen per cent of referrals (15/97) had no prior CT. Referrals who had a prior CT were triaged sooner than referrals without (1.7±1.6 months vs 3.9±3.3 months, p=0.013, 95% CI 0.48 to 2.94). Investigations and treatments received prior to respirology referral are presented in figure 1B.
Chart review of patients with ILD. (1A) Timeline depicting the average patient’s journey from symptom onset to ILD diagnosis. First CT scan occurred at 5.0 (12.1) months after first reported symptom. After the initial CT scan, submitting a referral to a respirologist took 2.4 (21.2) months. Patients were seen by respirology on average 10.5 (19.0) months from symptom onset, and diagnosed with ILD 12.0 (20.5) months from symptom onset. (A) First reported symptoms and (B) reason for referral to respirology are listed. (C) Bar graph figure depicts the length of time between symptom onset and diagnosis in relation to when a CT scan was completed. Patients who had earlier CT scans received an earlier ILD diagnosis. Note: values represented are median (IQR). (1B) Timeline depicting the average patient’s journey from first general practitioner visit to final ILD diagnosis. Patients who had a CT scan completed prior to respirology referral were triaged on a more urgent basis than patients without a prior CT scan (p=0.013). (A) Investigations and (B) treatments completed prior to respirology referral are listed. (C) Investigations ordered by the respirologist at the first patient visit. 6MWD, six minute walk distance; ABG, arterial blood gas; CT, computer tomography; DLCO, diffusion capacity of lung for carbon monoxide; FVC, forced vital capacity; ILD, interstitial lung disease; PFT, pulmonary function test. *10% of patients were asymptomatic at time of first respirology visit. ∼defined as either long acting muscarinic antagonist or long acting beta agonist. ˆdefined as combination of either LAMA, LABA or inhaled corticosteroids. #defined as azathioprine, mycophenolate or cyclophosphamide. Other investigations = sleep study, genetic testing, bone marrow biopsy, cardiology referral, gastroenterology referral, pH study, esophageal manometry, ventilation perfusion scan.
Table 1
|
Patient chart review and survey results
The most common reasons for respirology referral were ‘suspected ILD’ (62%) and ‘abnormal imaging’ (21%) (figure 1A). Once referred, patients waited 3.6±3.2 months to be seen. Respirologists commonly ordered 6 min walk test (58%) and serology (53%) at the initial visit (figure 1B). Five per cent of referral diagnoses were changed to a non-ILD diagnosis. The median time between symptom onset to final ILD diagnosis was 12.0 (IQR 20.5) months. Forty-six per cent (45/97) of patients were diagnosed within 1 year of symptom onset.
Patient survey
Of invited patients, 70/310 (23%) participated; mean age was 69.8±8.1 years, 50% were male. Initial patient-reported symptoms are listed in table 1. The time between symptom onset and seeking medical attention ranged from <3 months (41%), 3–6 months (17%), 6–12 months (9%) to >12 months (33%); 16% felt they waited too long to see their doctor. Patients reported their general practitioner ordered tests (57%), referred to a specialist (43%) and diagnosed and treated another condition (45%) (9% chronic obstructive pulmonary disease (COPD), 7% asthma, 6% bronchitis, 3% heart disease and 20% other). Eleven per cent were diagnosed with ILD by their general practitioner. Fifty-four per cent reported receiving no treatment, while 31% were prescribed inhalers, 16% antibiotics and 7% prednisone. Patient-reported time from symptom onset to an ILD diagnosis was <1 year for 51% and >1 year for 49%. Challenges in receiving a timely ILD diagnosis were experienced by 79% of patients (table 1).
Physician survey
Of invited physicians, 20/73 (27%) participated; mean time in practice was 16.3 years, practice location was 28% community and 72% academic. Fifty-five per cent of physicians had a unique approach to triaging ILD referrals; 73% saw patients more urgently, 73% ordered PFTs beforehand and 36% ordered CT scans beforehand. Challenges in providing a timely diagnosis were reported by 9/20 (45%) of respondents. The most commonly cited challenge was timely CT access (100%, 9/9). CT wait-times were <1 month for 25%, 1–6 months for 50% and 6–12 months for 25% of respondents. Other identified challenges included alternative initial diagnosis (67%), patients’ perception of their symptoms (58%), timely rheumatology consults (25%) and timely serologic results (25%). Physician-perceived gaps in ILD care centred around oxygen access (approval criteria, inadequate delivery symptoms and complex renewal processes), medication access (application complexity and reapplication procedures) and limited access to palliative care and pulmonary rehabilitation (table 2). Provincial ILD resources were frequently utilised (online supplemental table 2).
Table 2
|
Perceived gaps in ILD care and proposed solutions
Patient focus groups and thematic analysis
Thirty of 70 (43%) invited patients joined one of three focus groups, comprised of 6–10 patients each. Four major themes were identified (figure 2A).
Thematic analysis of patient and physician focus groups. (A) Thematic analysis of patient focus groups. (B) Thematic analysis of physician focus groups.
ILD awareness
Patients shared their experiences on how well known the signs, symptoms and diagnosis of ILD are within the medical community. Some described experiences where they presented with unrecognised hypoxemia and were not adequately evaluated. A patient with a nursing background explained:
I did not know what ILD was. If I had taken an O2 saturation and then just waited a few minutes and it came back up I would’ve thought, oh well, they're okay, not recognizing that that [could be] a whole other diagnosis.
Incidental radiographic ILD where patients had no dedicated follow-up was a commonly shared experience. Patients felt these instances were a missed opportunity for an earlier diagnosis.
I was kind of taken aback…[the doctor] says in 2011…I had a CT scan completed and the radiologist said there was ILD. And here we are almost 10 years later.
Pathway to diagnosis
Patient-related factors included patients delaying their presentation to healthcare as they attributed their symptoms as non-specific or due to another cause.
System and care provider-related factors were multifactorial. Patients described having to self-advocate to access tests, improve their wait-times or get information. A common perception was the need to understand how the healthcare system works and the need for self-advocacy. Patients commented on how patients and care providers may have different perceptions about the urgency of their presentation. Some patients felt the need to utilise private healthcare to expedite their diagnosis. In addition, patients felt that access to specialists varied greatly by geographical location:
I’ve lived in many places around Alberta, and I can say that without a doubt there are barriers to rural medicine. Quite often you have doctors that are in and out, or they're a locum. And they don't really take any deep interest. [ILD] is not a black and white issue…there’s 200 shades of grey.
The impact of comorbidities was felt to influence a patient’s experience as they were often initially referred for a non-pulmonary work-up. At times, they felt their symptoms were minimised leading to delays.
Impact of ILD diagnosis and living with ILD
Patients often recalled a clear diagnosis date, after a lengthy process of tests and appointments. Once given a diagnosis, patients described experiencing confusion, uncertainty and wished for more information. At times this was accompanied by an uncertain prognosis.
I asked a couple of questions about the implications of fibrosis which [the doctor] answered by saying, well nobody really knows. You could live a long life and die of something else, or it could … take you in three months, and so that was [not] very reassuring…
Patient and caregiver mental health was felt to be under-recognised and unaddressed.
Patient-informed improvement suggestions
Patients advocated for improved ILD medical training. They were encouraged by ongoing research in the field. Work-flow recommendations included faster time to referral and improved test access before a specialist appointment. Given the risk of irreversible progression, patients advocated for urgent triaging and ILD screening opportunities.
Physician focus groups
Twelve of 20 (60%) invitees participated in one of two focus groups, each comprised of six respirologists. Three major themes were identified (figure 2B).
ILD awareness
Respirologists described varied referral patterns from primary care. Common respiratory diagnoses, such as asthma and COPD, and treatment plans were often delivered to patients prior to respirology referral. The reason for referral and differential diagnoses could be vague. As most respirologists triage ILD patients more urgently, if ILD was not specifically mentioned, wait-times were longer and ILD-specific tests were not ordered.
Regional variations in radiologist expertise influenced time to diagnosis. Some respirologists sought a second opinion from a dedicated cardiothoracic radiologist.
Pathway to diagnosis and treatment
Physicians had varied definitions of appropriate triage times for suspected ILD, but generally aimed to see patients within 3 months of referral.
Diagnostic tests had varying wait-times. PFTs and bloodwork were fairly accessible, but timely CT access was a major barrier in some regions, with minimum understanding of why wait-times were so long for those experiencing them. This has led some patients to pursue private-pay options. The diagnostic complexity, including investigations and subspecialty referrals, lengthens time to diagnosis.
Once diagnosed, physicians reported that the time and resources required to care for ILD patients is higher than for most other conditions, ultimately impacting the care of other patients within a practice. For example:
I double block the patient, but that might be one less patient being seen during the day… Sometimes you wish for no-shows on that day for completing the work for (ILD patients)…if there is a full clinic …I tend to do (paperwork) at the end of the day, but it’s quite challenging. Even reviewing the declined medications or all the things that you need in order to get oxygen, you name it, it is time-consuming.”
Alternative models of care were thought to be beneficial. Respirologists remarked that the first appointment should be done in-person, but follow-up telemedicine appointments can improve timely care. ILD centres were utilised for a variety of reasons like diagnostic or management decisions, a surgical lung biopsy opinion and access to their allied healthcare teams.
Despite the benefits of ILD centres, physicians expressed frustration with wait-times and on occasion have avoided referring patients due to this. Community-based respirologists have often sought alternative forms of consultation, such as directly contacting the ILD physician outside of formal referral pathways or presenting cases themselves at MDDs. Access to MDDs varied by region, but were felt to be a useful resource and often replaced the need for formal ILD consultation.
Physician-informed improvement suggestions
Physicians advocated for shared-care models with general practitioners to improve patient care. However, in under-serviced regions, limited access to family medicine was seen as a barrier to this approach.
Physicians suggested that standardised referral criteria with recommended triage wait-times would be helpful. In addition, real-time knowledge of the ILD centres wait-lists would help facilitate shared care. Physicians reiterated the need for timely radiology tests and largely agreed that if general practitioners were comfortable ordering high resolution CTs this would expedite diagnosis. Finally, physicians felt that webinars and ILD colleagues’ advice were most useful to remain up-to-date on ILD knowledge, while guidelines were less utilised.
Discussion
This mixed-methods study highlights that a timely diagnosis of ILD may be impacted by patient-related, healthcare provider-related and system-related factors. Patients and physicians highlighted important opportunities to improve care for ILD patients, including earlier presentation to primary care and access to timely diagnostic tests and specialist care.
There is no gold standard time-to-ILD-diagnosis. Although earlier diagnosis is advocated, delays are perceived across all stages of a patient’s journey.7 A typical experience is characterised by misdiagnosis, exposure to costly and invasive diagnostic procedures and substantial use of healthcare resources.8 9 Previous literature has suggested that ILD diagnostic delays occur early in a patient’s journey, prior to assessment by a specialist.10 Our data mirrored these results with patients experiencing symptoms for a median time of 5 months prior to their initial CT scan and 12 months prior to their final ILD diagnosis. Our data support that a delayed diagnosis is concordant with a delay in time to first CT. The reasons for these delays are multifactorial. The minority of our patient cohort (16%) felt that they delayed presenting to their general practitioner. Conversely, 51% of patients felt that their healthcare providers did not recognise their lung problem early. This may be unsurprising given ILD’s non-specific symptoms, which may be attributed to ageing, smoking or more common cardiorespiratory diseases.9 10 A large US-based study of 600 ILD patients reported misdiagnoses occurring once (55%) or twice (38%) prior to their ILD diagnosis. Subsequently receiving treatment for these more common presentations can further impact the time-to-diagnosis.5
Another possible source of diagnostic delay is the under-recognition and under-reporting of ILD features on CT,7 particularly when chest imaging is conducted for other indications. A recent systematic review and meta-analysis reported an 11% prevalence of interstitial lung abnormalities (ILAs) in a pooled cohort of lung cancer screening, general population and at-risk familial cohorts.11 While not all ILAs will progress to clinical ILD, this high volume of at-risk individuals informs the need to develop standardised CT reporting and follow-up, especially given ILAs are a risk factor for mortality.12 As per the informed suggestions of our patients and physicians, we advocate for early respirology referral when a CT reports features of ILD. In addition, many respirologists sought second opinions from dedicated cardiothoracic radiologists, which also may have had an impact on time to diagnosis. This could be rectified by increased radiology training in ILD or potentially the implementation of artificial intelligence as a means to improve diagnosis efficiency and accuracy.
Our work suggests that strategies targeting various points and stakeholders in a patient’s diagnostic pathway may reduce delays. Education and awareness of ILD in primary care may increase recognition of this serious disease. Implementation of novel screening algorithms embedded within electronic medical records (EMR) may help identify ILD using readily available information such as diagnostic codes, age, sex and smoking history. Such algorithms can identify patients with IPF before they present clinically,13 leverage readily available clinical data, require no invasive testing and have minimal upfront costs or provider training. EMR-based diagnostic checklists may also improve diagnostic processes, ensuring standard approaches to care.14 Partnering with radiology stakeholders can help streamline CT follow-up by adapting the care pathway already established for suspected chest malignancies. Radiologists flag incidental findings and prompt an appropriate referral. This could be tailored for parenchymal lung changes to help expedite respirology referrals. Increasing access to virtual MDDs may relieve ILD clinic wait-times, as could increasing the number of ILD experts and centres to provide more equitable access to subspecialty care.15
This study has limitations. Chart reviews, surveys and focus groups were restricted to respirologists, and may not be generalisable. Subsequent work should explore the experiences of general practitioners managing undifferentiated dyspnoea and barriers to a timely ILD diagnosis within primary care. The sample size was underpowered to analyse time to diagnosis by ILD diagnosis. It is possible patients with pre-existing autoimmune disease may have a higher clinical suspicion for ILD and therefore be diagnosed earlier. Survey patients were recruited from ILD support groups, which are often linked to ILD centres and their experiences may not be generalisable to community patients. It is possible this created selection bias and these engaged patients may have sought care earlier and the barriers identified are even more accentuated in less empowered patients. Survey and focus group data were subject to recall bias, as evidenced by the difference in reported prior treatments between physicians and patients. In addition, interviewers had knowledge of healthcare delivery and while every effort was made to remain neutral during interviews, this may have influenced dialogue. Finally, our study represents experiences within the Canadian healthcare system and may not be generalizable.
Conclusions
This study identified several sources that led to the diagnostic delay of patients with ILD. Advocating for increased awareness, improved access and report standardisation of CT scans, and creation of ILD centres may improve time-to-diagnosis. Optimising the diagnostic journey of patients will require engagement of multidisciplinary stakeholders, development of local clinical practice recommendations and the implementation of evidence-based diagnostic tools and processes aimed at improving the accuracy and reducing the inherent diagnostic complexity of ILD.