The number of studies relating to teledermatology continues to rise exponentially. The reasons for this are multitude, including the COVID-19 pandemic limiting in-person consults, advances in technology meaning better availability of telemedicine and, in particular, teledermoscopy, as well as more familiarity and interest from health care professionals in using teledermatology. There are various methods of teledermatology. The main distinguishing factor is whether a pathway uses video consultations or store-and-forward, also known as advice and guidance. The former uses real-time technology to provide assessment, while the latter involves taking images for later review. Teledermatology can also be distinguished by the specialties involved, whether this is the more common general practitioner (GP) to a dermatologist, dermatologist to dermatologist, GP to GP, or patient to GP.

There are many variables to consider when assessing store-and-forward teledermatology due to the variation in the way it is implemented and assessed. This review aims to collate information from the breadth of available data.

This literature review discusses time to advice, effectiveness, and image quality for store-and-forward teledermatology for skin lesion assessment.

In total, 11 papers reporting time outcomes were identified with measures including time to advice, time to biopsy, and time to definitive surgery. The results are outlined in Table 1.

In total, 34 papers assessing the effectiveness of teledermatology assessment were identified. These included accuracy and its derivations, concordance, positive and negative predictive values, and impact on subsequent in-person assessment. The results are outlined in Table 2.

In total, 8 papers about image quality in store-and-forward teledermatology were identified. The results are outlined in Table 3.

The majority of studies assessing store-and-forward teledermatology are observational studies. Of the small number of interventional studies identified, only 2 were randomized controlled trials. The area with the most evidence available is for effectiveness, with 34 out of 45 studies reporting various outcomes of efficacy. Overall, the majority of studies in this literature review report that store-and-forward teledermatology services allow reduced time to advice, comparable effectiveness to in-person assessment, and at least adequate image quality for most skin lesions.

One of the main advantages of teledermatology in skin lesion analysis is the potential for expedited treatment through the reduction of time between referrals and management. There have been several studies assessing this question. They predominantly show improvement with reduction ranging between 4 and 70 days for outcomes such as biopsy, definitive treatment, or clinic appointment [1-6]. Moreno-Ramírez and Ferrándiz [7] report reviewed 43,677 patients over 10 years with an average time to advice of 72 hours. Only 2 studies showed no difference in time to biopsy or definitive treatment [8,9]. A study by Duong et al [8] was underpowered with only 36 patients in the teledermatology pathway and compared in-person consults from 3 to 5 years earlier when there was potentially lower service demand. Very little detail on the teledermatology process used by Creighton-Smith et al [9] is provided. Perhaps, the lack of time reduction is a product of their teledermatology pathway design in this single-center study.

A retrospective review of 4589 referrals by Lowe et al [10] showed that using community imaging for skin lesion referrals enabled 86.3% of referrals to be managed without subsequent in-person assessment. A systematic review of 21 studies by Finnane et al [11] concluded that teledermatology could allow reduced wait times and earlier skin lesion management based on variable reduction in waiting time in 3 studies. Overall, there is a significant time-saving benefit when using teledermatology to review skin lesions.

The effectiveness of teledermatology is difficult to evaluate. One of the most prominent reasons is the heterogeneity in teledermatology services, making direct comparisons difficult. There is variable use of teledermoscopy and the provider experience in teledermoscopy. The study designs are often small, single-center studies, and the nature of teledermatology means that significant selection bias is often at play. The existing interventional studies often limit generalizability to clinical use by recruiting from settings other than primary care and have no way of allowing blinding. Another issue is the gold standard in diagnosis. Ideally, this would be a histological diagnosis; however, this is usually only available in a small proportion of the most suspicious lesions. In the absence of histology, often the reference standard defaults to in-person dermatologist assessment, which biases the diagnostic effectiveness away from teledermatology assessment. Despite this, a Cochrane review by Chuchu et al [12] focusing on diagnostic accuracy in teledermatology for skin lesion assessment concluded that teledermatology was accurate enough to diagnose most malignant lesions. This was a review of 22 studies published up to August 2016. Their estimate of the overall sensitivity was 94.9% (95% CI 90.1-97.4), and specificity 84.3% (95% CI 48.5-96.8) [12]. Even with their concerns regarding the quality of the studies, they recommended that teledermatology be used to triage patients requiring in-person assessment.

Overall diagnostic accuracy of teledermatology in assessing skin lesions is high. Multiple studies have shown it to be similar to in-person reviews, with figures ranging from 79.5% to 100% [13-17]. Often, this consists of high sensitivity and relatively lower specificity [18-21]. One study, focusing on 321 nonpigmented lesions for consideration of Mohs micrographic surgery, found that despite the high sensitivity of 92%, the specificity was only 49% [22]. van der Heijden et al [23] reported a lower accuracy, with only moderate teledermatology and histology agreement on 108 lesions, likely due to a high proportion of poor-quality images. The trend of high accuracy is consistent with a systematic review published by Finnane et al [11], which included studies from 2009 to 2015. This showed high diagnostic accuracy with the use of teledermatology. It was slightly inferior to in-person assessment at 51%-85% compared to 67%-85% [11]. The generally higher values of accuracy and sensitivity seen in studies from more recent years may reflect improvements in technology and greater experience with teledermatology diagnosis. Two studies looked specifically at positive predictive value when diagnosing melanoma [24,25]. They obtained values of 38.1% and 13.7%. The difference may reflect variability in dermatologist assessment with 9 readers, and significant variability within these noted in the latter study.

Similar to accuracy, there is a high degree of variation in diagnostic concordance reported in skin lesion teledermatology research. Overall, most studies indicate a high diagnostic concordance level compared to in-person dermatologist reviews and histopathology. A well-designed, repeated measures study by Warshaw et al [26] compared dermatologist and teledermatologist diagnoses of 3021 skin lesions from primary care. They report fair to substantial agreement for primary diagnosis of 45.7%-80.1% (κ=0.32-0.62) and substantial to perfect agreement for aggregated diagnoses (a primary diagnosis and up to 2 differential diagnoses) of 78.6%-93.9% (κ=0.77-0.9) [26]. Creighton-Smith et al [9] showed improved rates of partial concordance, and no difference in perfect concordance for teledermatology diagnoses compared to in-person dermatology diagnoses at 52% versus 36.3% (P<.05) and 36% versus 43.1% (P=.4), respectively. Dahlén Gyllencreutz et al [27] reported high interobserver agreement in skin lesion diagnosis between 2 dermatologists reviewing 172 images from primary care (81.4% and 83.7%). Zink et al [28], Arzberger et al [29], Lamel et al [30], Carter et al [5], and Kroemer et al [20] reported high diagnostic concordance for teledermatology compared to in-person or histology. In contrast, a small study of 10 lesions showed higher concordance for in-person review than teledermatology (κ=0.6 vs 0.52) [31]. Greenwald et al [32] conducted a real-world study assessing the histological diagnosis of pigmented lesions excised on the advice of a store-and-forward teledermatology service. In total, 260 of 1572 lesions (17%) were found to be melanoma, leading to a benign:malignant ratio of 4.9:1 [32]. This reduced with increased dermatologist diagnostic confidence to 2.8:1 for “excise, possible melanoma” and 1:1 for “excise, likely melanoma” when compared to the default “excise to remove doubt.”

A benefit of teledermatology is the reduction in patients requiring in-person assessment. A retrospective review of 59,279 primary care referrals found a 39% decrease in face-to-face appointments using the teledermatology pathway [33]. Mehrtens et al [34] found that teledermatology allowed 50% of referrals to be discharged to GPs and 33% to proceed straight to biopsy, saving an estimated 16,282 in-person appointments. Several other studies also noted substantial reductions in face-to-face appointments after implementing teledermatology services [13,35,36].

One of the most important distinctions when reviewing skin lesion assessment by teledermatology is the use of teledermoscopy. This is particularly important when pigmented lesions are assessed. Most research suggests adding dermoscopy images for teledermatology is more effective than clinical images alone. Reviews by Beer et al [37] and Woodley [38] concluded that teledermoscopy increases lesion diagnosis accuracy. Gómez Arias et al [39] compared diagnostic concordance with and without teledermoscopy in 395 cases and showed increased concordance when dermoscopy was included. Şenel et al [40] reported improved accuracy in diagnosing 150 patients with nonpigmented lesions with dermoscopy compared to without (0.86 and 0.88 vs 0.75 and 0.77; P<.05 for 2 different dermatologists). Several studies have reported the high effectiveness of teledermatology when using dermoscopic images [24,41,42].

In contrast, Janda et al [43] and de Giorgi et al [31] found no increase when dermoscopy images were added to teledermatology assessments. In the former, images were acquired directly from 234 participants through skin self-examination. Images were likely to be particularly poor if participants were unaware of the need to optimize the skin surface before obtaining dermoscopy imaging. The latter was a small trial of 10 lesions, and as the authors note, there were many difficult lesions.

The availability of clinical history is another factor that may impact teledermatology diagnosis. Surprisingly, it is difficult to find details on this area within studies, making the effect of this difficult to determine.

Image quality is an essential aspect of skin lesion teledermatology. Many studies have reported the proportion of images adequate for diagnosis. These range from 0.2% to 36% [13,18,20,23,36,45]. Dahlén Gyllencreutz et al [27] compared image quality from primary care practices using iPhone 4 cameras with dermoscopy attachments to those taken by dermatologists with Canon EOS D550 cameras. In total, 172 cases were reviewed by 2 dermatologists not familiar with the cases. Despite the differences in camera resolution and operator experience with dermoscopy, there was no statistical difference in the quality of images obtained. The evaluators rated 1.7% and 4.7% of primary care images as poor compared to 1.2% and 3.5% of dermatology images (P=.25 and P=.28) [27]. They postulate that the difference between evaluators reflects the subjective nature of image quality assessment. Vestergaard et al [18] also had individual dermatologists review the quality of 600 images from primary care. Unlike Dahlén Gyllencreutz et al [27], Vestergard et al [18] found substantial agreement in image quality, with approximately 10% reported as poor quality by each reviewer.

Image quality in teledermatology depends on several factors, such as the equipment used, the inclusion of dermoscopy, and the experience of the person taking images. Studies have shown images obtained by dermatologists or dermatology nurses to be of higher quality than those from primary care [23]. Images obtained by patients generally have the lowest image quality [45]. This is likely due to the increased difficulty taking the images, less experience in focusing lesions, and in the case of dermoscopy, unfamiliarity with the need to prepare the skin, such as removing make-up or applying a fluid to the lesion. Given advances in imaging quality, notably improved smartphone camera resolution, more recent research is likely to report higher image quality regardless of the operator.

Overall, teledermatology offers a comparable standard of effectiveness to in-person assessment. It can save significant time in expediting advice and management. Image quality and inclusion of dermoscopy have a considerable bearing on the overall effectiveness. There is a need for large interventional studies, particularly those with high proportions of histology available to enable definitive conclusions regarding teledermatology outcomes. There is a gap in the literature for studies comparing different teledermatology methods.