The MEL-SELF pilot trial ran from November 2018 to January 2020, and a detailed report of the study findings has been published [11]. A total of 100 patients previously treated for melanoma were recruited from specialist-led and GP-led private clinics in Sydney and Newcastle, New South Wales, Australia, and randomized to control (n=51, 51%) or intervention (n=49, 49%). Three treating dermatologists and 6 skin specialist GPs recruited patients, provided routine skin checks, and reviewed lesions as prompted by the intervention, while images submitted from patients by teledermatology were sent to and reviewed by 3 dermatologists located in Sydney, Brisbane, and Melbourne (1 of whom also recruited patients and provided their ongoing care). The model of teledermatology in the MEL-SELF pilot trial was one in which the teledermatologist was not the patient’s treating doctor. Treating doctors did not assess images submitted by teledermatology.

Patients in the intervention group were guided on skin self-examination (SSE) through online videos [26] and were provided with a mobile dermatoscope and smartphone app. Patients took dermoscopic images and submitted these for teledermatology review, together with self-reported history (ie, lesion location, history of change, melanoma history of themselves and family, number of moles, skin type, and patient age). If the teledermatologist assessed the lesion as suspicious for melanoma and recommended urgent clinical review in their report, the patient made a fast-tracked, unscheduled appointment with their treating doctor (facilitated by research staff). Teledermatology technologies were provided by MetaOptima Technology Inc (Vancouver, Canada) [27], including a mobile dermatoscope (MoleScope I) that integrates with MoleScope (a smartphone-based skin imaging app) [28] and DermEngine (a digital software system that facilitates the capture, storage, communication, and analysis of skin images by dermatologists) [29].

Clinicians were eligible to participate in this qualitative study if they were involved in screening and recruiting patients for the MEL-SELF pilot RCT and/or reading submitted images and providing reports via teledermatology.

This study was approved by the Human Research Ethics Committees at the University of Sydney (X15-0445) and the Royal Prince Alfred Hospital (HREC/15/RPAH/593). All participants provided informed consent. The design, conduct, and reporting of this study follow the SRQR (Standards for Reporting Qualitative Research) guidelines [30].

An interview topic guide was developed by the authors to cover views on and experiences of patient-led melanoma surveillance, teledermatology, patient-performed teledermoscopy, and components of trial implementation relevant to each clinician depending on the role they played (Multimedia Appendix 1). The topic guide was flexible enough to adapt to the specific role of each clinician in the pilot trial, and discussions beyond their experiences in the pilot trial were also encouraged. Semistructured phone interviews were conducted by 1 researcher (author DD) who is trained in qualitative interviewing and was not known to any of the clinicians before the study. Interviews lasted between 19 and 77 minutes, with a median time of 23.5 minutes. Interviews were audio recorded and transcribed verbatim by 2 researchers (authors DD and EH). Identifying details were removed from the data at the transcription stage.

Interview summaries were written immediately following each interview. Discussions between authors DD and EH considered emerging analytical ideas and opportunities to refine the topic guide and approach. Preliminary codes were developed inductively from the interview summaries by DD and EH. Both authors tested these by separately coding 4 transcripts. A coding framework was then agreed upon by DD and EH and applied to the rest of the data. Key themes and interpretations were identified through coding, memo writing, and analytical discussions [31] involving authors DD, EH, JH, and KJLB. Our emerging theoretical framework had clear resonance with 2 concepts pertinent to the assessment, improvement, and prioritization of health care service delivery: high-value care and medical overuse. High-value care refers to necessary health care that is supported by evidence showing it benefits patients [32]. Conversely, medical overuse is unnecessary health care that is unlikely to benefit patients but may cause them harm [33,34]. We used these concepts to further organize and interpret the data. Coding and initial organization of codes were done in NVivo 12 software (QSR International), and a data matrix was exported to Microsoft Excel (IBM Corp) to aid thematic analysis.

All 11 clinicians who were involved in the MEL-SELF trial were invited to participate in an interview. Three clinicians did not reply to the initial or follow-up invitation email, while 8 clinicians agreed to participate and were interviewed between October and November 2020. These included 6 treating clinicians (2 dermatologists, 1 of whom also provided teledermatology, along with 1 surgical oncologist and 3 skin specialist GPs) and 2 dermatologists who provided teledermatology only. Five clinicians were female and 3 were male. Clinicians’ experience in managing melanoma patients ranged from 15 to 20 years.

All clinicians had experience using teledermatology; however, due to the novelty of patient-led teledermoscopy in the MEL-SELF pilot trial, experience reviewing dermoscopic images taken by patients (especially those who were not under their care) was more limited. Only 1 clinician (a teledermatologist) had extensive experience reporting on dermoscopic images taken by patients in research settings. All had experience providing second-opinion reviews of dermoscopic images taken by clinical staff in both clinical and research settings. Along with their experiences in the pilot trial, clinicians discussed their experiences of teledermatology outside of it, and these accounts were included in the analysis.

The results of the thematic analysis are summarized in Figure 1. For patient-conducted teledermoscopy to deliver high-value care, strategies are needed to both facilitate potential benefits and inhibit potential harms (particularly from medical overuse). Themes and their relationships to each other are explained in the subsequent sections, supported by illustrative quotes. Additional quotes are included in Multimedia Appendix 2.

There was an agreement among clinicians that patient-led surveillance is a good idea. Clinicians said that patients should be encouraged to get to know their skin and be educated in self-monitoring, noting, however, that not all patients are interested in doing so. Teledermatology was thought to increase access to dermatology services for rural and remote patients and enable continuity of care for all patients during the COVID-19 pandemic, thereby facilitating the early detection of melanomas. Teledermoscopy was thought to make shared monitoring of particular lesions of concern more convenient for high-risk patients living at a distance from dermatology services. Clinicians said that teledermoscopy enables quick feedback for lesions that the patient thinks are potentially concerning, either facilitating prompt review in the clinic if necessary or otherwise alleviating worry and a potentially long period of anxiety ahead of the next routinely scheduled visit. They also thought it may be useful for monitoring lesions that the clinician thinks are potentially concerning. For this use, remote monitoring was thought to have beneficial potential by reducing the frequency of routinely scheduled clinics for monitoring particular lesions.

This qualitative interview study provides important insights into the experiences and perspectives of a highly specialized group of clinicians (dermatologists and GPs specializing in skin cancer care) who participated in a pilot RCT of patient-led melanoma surveillance using mobile teledermoscopy. Clinicians identified several potential benefits of such an intervention, including additional monitoring and early diagnosis, reassurance for patients, increased access to care, and a reduction in unnecessary clinic visits. Clinicians also discussed several challenges that may increase clinical uncertainty and lead to medical overuse, as well as potentially harming patients. These included receipt of low-quality images from patients (also highlighted by Kozera et al [35]), limitations to assessment of lesions from a dermoscopic image, potential for the patients to miss lesions of concern, and perceived medicolegal risks.

We found that to decrease the potential for medical overuse when implementing patient-led surveillance using teledermoscopy, it is necessary to manage the clinicians’ uncertainty. Uncertainty could be reduced by providing teledermatologists with adequate clinical history, offering the intervention to patients who have been screened for the capacity to use app and photo functions of their smartphone with ease and have someone to help them take dermoscopic photos, and providing patients with sufficient training and ongoing support. For the potential benefits of the intervention to be realized, it is important to ensure its optimum placement within the clinical care pathway. Clinicians suggested that the intervention may be able to replace scheduled monitoring visits for some patients and that the intervention could be used as a triage tool to decide whether in-person assessment is necessary for new lesions. These strategies may ensure that the intervention is most likely to deliver benefits without causing harm.

Although there is potential for the intervention to deliver high-value care, during the pilot trial, several interacting factors resulted in some instances of medical overuse, including unnecessary unscheduled visits and biopsies. Our findings suggest that sometimes teledermatologists (who, as per the trial design, were separate from the treating clinicians) recommended that patients make a fast-tracked unscheduled visit to review lesions that they considered suspicious because they did not have enough information about the lesion or the patient or the image was of poor quality. These decisions may reflect risk aversion, intolerance of uncertainty, and fear of malpractice and litigation, factors that may drive medical overuse [36,37]. In these instances, the treating clinician may have ordered unnecessary biopsies that they otherwise would not have to reassure the patient following an alarming teledermatology report. It is also well known that patients experience significant anxiety concerning their melanoma diagnosis [38], so perceived patient desire for clinical action by clinicians wishing to relieve patient anxiety and prioritize maintenance of the doctor-patient relationship could result in decisions to undertake biopsies even if they think this is clinically unnecessary [39]. Some of these issues may be avoided if patient-led surveillance is implemented with the patient’s treating doctor as the teledermatologist. Where this is not possible, sufficient clinical information needs to be provided to the teledermatolgist.

Underlying the cautious approach taken by teledermatologists and treating clinicians is that patients who participated in the MEL-SELF trial had a personal history of melanoma and were thus at high risk of a subsequent melanoma. The framework by Greenhalgh et al [40] for considering influences on the adoption, scale-up, spread, and sustainability of patient-facing health care technologies considers the nature of the condition as the first domain in theorizing the success or failure of interventions. The diagrammatic depiction of our findings (Figure 1) suggests the ever-present influence of the high-risk nature of melanoma surveillance on clinicians’ perceptions and decisions related to mobile teledermoscopy due to the potential consequences of a missed (or delayed) diagnosis. It follows then that perceived medicolegal risk was high among our sample of clinicians and has been similarly reported in other studies [41], even though perceived risk may be higher than actual risk [42]. Our findings suggest that perceived consequences (for patients and clinicians) of a delayed diagnosis, including medicolegal risk, may have impacted the provision of care. This issue needs to be addressed to minimize the potential for medical overuse from patient-conducted teledermoscopy. Possible solutions might include educating clinicians on the lower-than-perceived actual medicolegal risk, the potential for medicolegal action arising from harm from medical overuse, and the need for transparency about the uncertainty of teledermatology assessment when discussing the process with patients.

Potential harms to the patient from medical overuse such as anxiety, risk of complications from medical procedures, changes to physical appearance, and effects of disease labeling [43] should also be considered by the doctor when making a recommendation via teledermatology or deciding to do a biopsy. Clinicians’ uncertainty associated with patient-performed teledermoscopy may possibly lessen over time as they gain more experience and confidence using the new technologies. A “transition period” after the introduction of new technology, in which clinical management thresholds change and then return to resemble what they were before its introduction, is well documented [44].

This study is one of the first to explore the experiences of specialized clinicians involved in patient-led surveillance using teledermoscopy. A strength of our study is that our findings are based on unique perspectives from clinicians who had different roles in delivering the patient-led surveillance intervention. Their views reflected their experiences during the MEL-SELF pilot trial, other trials of patient-performed teledermoscopy, and teledermatology in their practice in general. However, accounts of mobile teledermoscopy in a trial context may differ from experiences in clinical practice. In particular, unlike in our trial, the treating clinician may often be the teledermatologist in clinical practice. We also acknowledge that our study sample was small, and our findings are not intended to represent the breadth of experiences and views within each clinician group, namely, skin specialist general practitioners, dermatologists and teledermatologists, and surgical oncologists. A larger sample may have also revealed differences in perspective and opinion between the clinician groups that we did not detect. Additionally, the scope of our inquiry did not allow us to pursue all factors that were previously been found to impact clinicians’ views of teledermatology such as reimbursement, patient privacy, and internet security [7,45]. Finally, the findings of our study are influenced by the health system context in which the clinicians work. Although in Australia, community-based health care receives public subsidy through the Medicare Benefits Scheme, there are still significant out-of-pocket costs, on average AU $183 (US$124) over 6 months for patients in the MEL-SELF pilot trial [46]. All clinics involved in the study are privately operated and charge a fee for service. Moreover, competition between private melanoma clinics exists, but this may not be comparable to the provision of melanoma care in other contexts such as the United States where a broader system of universal health care is not present.

In response to this study’s findings, we made improvements to trial processes for the larger ongoing MEL-SELF trial (ANZCTR12621000176864) [19]. Teledermatologists now have access to an image of the patient’s back to give context on the patient’s skin in general (eg, signs of sun damage), and each patient selects a target lesion with their treating doctor, making explicit the shared monitoring of the lesion. To facilitate monitoring of a lesion, each patient is allocated to 1 teledermatologist who will review all the patient’s images and be able to assess changes over time. Teledermatologists also have the option of seeking a second opinion from another teledermatologist using the DermEngine platform. Each teledermatologist will receive information on subsequent clinical decisions and outcomes as feedback for lesions they have reported on. To encourage perception and use of the intervention as a triage tool, unnecessarily alarming and suggestive terms such as “urgent” or “biopsy” are no longer used in the teledermatologist’s feedback to patients. Further decisions regarding clinical intervention (eg, biopsies, excisions) are left to the treating doctor who reviews the lesion in-person, including the decision to override the teledermatologist’s recommendation. The ongoing MEL-SELF trial is also collecting information on the quality of images taken by patients and whether teledermatologists can report on them.

This study illuminated the potential benefits and challenges of using patient-performed teledermoscopy from the perspectives of clinicians involved in a pilot RCT of patient-led surveillance. Patient-led surveillance may address inequities in access to melanoma surveillance for patients who live remotely, are less mobile, or require continuity of care during a pandemic [47,48]. It has the potential to deliver high-value care, but safeguards are needed to mitigate against the potential for medical overuse [33,34,36]. The larger MEL-SELF trial, with refinements to the teledermoscopy process in response to the findings from this study, will provide robust evidence on clinical effectiveness to inform the potential wide-scale adoption of patient-led surveillance.