Unimaged Melanomas in Store-and-Forward Teledermatology

Introduction

Store-and-forward teledermatology (SFT) has gained wide acceptance as a means for delivering specialist care in a variety of clinical settings. ^{1 –5} While diagnostic accuracy and appropriateness of directed management plans have been shown to be noninferior to face-to-face (FTF) dermatology, ⁶ concerns remain regarding the safety of SFT in the diagnosis of malignant melanoma. ^{7 –11} This concern is more related to the perceived inability of primary care providers (PCPs) to direct imaging of suspicious lesions than to the SFT dermatologist's ability to diagnose them. ^{12 –14}

At least three previous studies have ascertained the frequency of “incidental” melanomas in populations of patients referred for FTF dermatology care. An incidental melanoma is defined as a melanoma discovered by a dermatologist during full body skin examination (FBSE), but was not the subject of the referral. Viola et al. discovered 6 incidental melanomas in 400 referrals for suspected skin cancer to the VHA Connecticut Healthcare System. ¹² Aldridge et al. discovered 5 incidental melanomas among 336 patients referred from primary care for evaluation of lesions suspicious for skin cancer. ¹³ Kingsley-Loso et al. discovered 87 incidental melanomas among 17,174 referrals to a VA dermatology clinic. ¹⁴ Based on these studies, there is justifiable concern that unimaged melanomas may be present in patients being referred to SFT for other dermatologic concerns. ^{1,12 –15}

We conducted this quality improvement study to ascertain the frequency at which unimaged melanomas occurred among patients enrolled in a VA SFT program and to determine what, if any, policy and procedural changes might be necessary to improve the program.

Materials and Methods

Results

During the period from July 1, 2009 through December 31, 2011, 12,863 SFT consultations were completed on 7,960 unique Veterans. Sixty-nine melanomas met inclusion and exclusion criteria. Fifty-six were the subject of the SFT consultation; 13 were unimaged melanomas. Using the total number of SFT consultations performed as the number of opportunities to properly image a melanoma, we calculated the frequency of unimaged melanomas in our population to be 13/12,863 (0.101%) or 10.1 per 10,000 consultations.

We conducted secondary analyses to characterize unimaged melanomas by melanoma location, discoverer, staging, and interval between SFT visit and biopsy (Table 1). We hypothesized that melanomas easily visible during a PCP visit would be less likely to be unimaged. We considered melanomas on the head, neck, hands, and forearms to be easily visible during a primary care visit. All other locations, such as back, trunk, or legs, were not considered easily visible. Using these criteria, we discovered that 24 of 56 (42.8%) imaged melanomas were easily visible to the provider. Four of 13 (30.7%) unimaged melanomas were easily visible to the provider (Table 1). We found no significant relationship between a melanoma not being easily visible and unimaged (Z score 0.882, P = 0.3779, 95% CI: 9.05 to 61.36).

We discovered that of the 13 unimaged melanomas, only two were discovered later by a dermatologist. The others were found by Veterans, non-M.D. dermatology practitioners, PCPs, or by SFT trained nurses.

Of the unimaged melanomas, 7 of 13 (53%) were invasive. In contrast, 36 of 56 (64.3%) imaged melanomas were invasive (Z score 0.850, P = 0.3951, 95% CI: 24.46, 80.15). Prior studies have shown that dermatologist discovered melanomas were thinner melanomas, while more invasive melanomas are found by the PCP. ¹⁴

We ascertained the time interval between the biopsy and the previous SFT visit. This ranged from as few as 21 days to as many as 930 days (Table 1). The average period of time that a melanoma was on the body was 502 days. Only six melanomas were discovered within 1 year of the SFT consultation. Seven were discovered within 2 years of the SFT consultation, and the remaining five were discovered more than 2 years after the SFT consultation.

Discussion

The principal findings of this quality improvement study were: (1) 13 unimaged melanomas occurred in 12,863 consultations, a frequency of 10.1 per 10,000 consultations. (2) Six unimaged melanomas were diagnosed within a year of a previous SFT visit. (3) Of the unimaged melanomas, six were pTis, five were pT1a, one was pT1b, and one was pT2b. (4) Only 4 of 13 (30.7%) unimaged melanomas would have been easily visible to a PCP.

Both SFT and FTF dermatology rely upon PCPs to refer patients for care. In both models, PCPs may fail to identify suspicious lesions. Melanomas found by an FTF dermatologist, but not initially identified by the referring PCP, have been termed “incidental melanomas.” ¹² Kingsley-Loso et al. posited that up to 37% of melanomas would have been missed without a FBSE. ¹⁴

Based on concerns raised by studies of “incidental melanomas,” we conducted this quality improvement study to ascertain the frequency of unimaged melanomas in our SFT program. Thirteen of 69 (18.8%) melanomas occurring in our patient population were unimaged. This frequency is likely an overestimate. Based on previously published melanoma incidence ¹⁵ (3.6 per 10,000), we would have expected a minimum of three melanomas to naturally arise in the study period for this Veteran population. This expectation is supported by our observation that five unimaged melanomas were diagnosed more than 2 years after the SFT consultation. There is a strong likelihood that these melanomas were not present on the Veteran during the PCP visit. For instance, Veteran #13 had a self-discovered melanoma 930 days, over 2.5 years, after the previous SFT encounter. It is very likely that this melanoma was not present at the time of imaging.

If SFT Veterans were lost to follow-up (LTFU), our results would underestimate the number of unimaged melanomas. To estimate the magnitude of the error, we selected a random sample of 453 Veterans who had at least one SFT consultation and ascertained whether they had received any VA care more than 1 year after the most recent SFT encounter during the period of observation. If they had received care, they were not considered LTFU. If their care ended within 365 days of the SFT visit, they were considered LTFU. Sixteen (3.5%) Veterans were found to be LTFU by this criterion. Accordingly, we would expect 281 Veterans of the entire SFT cohort to have been LTFU. The incidence of melanoma in Veterans in VISN 20 was reported to be 3.6/10,000. ¹⁵ Using this annual incidence, the number of incidental melanomas occurring among Veterans LTFU from the SFT cohort is 0.1.

We conducted secondary analyses of our unimaged melanomas to understand why they were not imaged. Six of our 13 unimaged melanomas were pTis, 5 were pT1a, 1 was pT1b, and 1 was pT2b. These results are similar to those of Kingsley-Loso et al. who found that 65% of their incidental melanomas were pTis. This suggests that PCPs are effective at recognizing and directing imaging of invasive melanomas, but may need additional training and experience to recognize the more subtle thin melanomas. Nine of 13 (69.3%) unimaged melanomas were not easily visible to the provider. These results are similar to those reported by Kingsley-Loso et al. who described 55 of 87 (63.2%) incidental melanomas as not easily visible. Veterans and PCPs must be made aware that melanomas can occur on not easily visible parts of the body and a directed educational program may be of value. These findings underscore the potential value of education for Veterans and PCPs regarding the limitation of SFT, thereby facilitating development of a partnership among Veterans, PCPs, and SFT dermatologists to improve diagnosis and management of malignant melanoma.

FBSE conducted by a dermatologist is the gold standard for discovering melanomas. Our SFT program, as currently implemented, does not match this standard. Due to a shortage of dermatologists in VISN 20, SFT remains the best option for delivery of specialty dermatology care in places where access to FTF dermatology is limited. Healthcare administration must recognize this limitation when implementing SFT programs. A promising avenue for decreasing the frequency of unimaged melanomas was suggested by Avilés-Izquierdo et al. who found that men over 70 years old are a logical target for melanoma detection education. ¹⁶ Structured programs or informational communications may increase patient awareness and encourage FBSE during PCP checkups.

Limitations

This study only included Veterans, and may not be generalizable to a wider population. Unimaged melanomas diagnosed by non-VA providers, and not represented on the problem list, would be missed by our analysis.