Diagnosis and Management of Malignant Melanoma in Store-and-Forward Teledermatology

Abstract

Background:

Published studies have led to concern that store-and-forward teledermatology (SFT) diagnosis and management of melanomas may be inferior to face-to-face (FTF) dermatology care.

Introduction:

To ascertain the frequency of correctly managed and diagnosed melanomas within a population of veterans in Veterans Integrated Service Network 20 SFT.

Materials and Methods:

We conducted a retrospective chart review of 7,960 veterans seen by SFT between July 1, 2009 and December 31, 2011.

Results:

Of the 61 veterans that met inclusion and exclusion criteria, 45 (74%) melanomas were correctly diagnosed and 57 (93%) were correctly managed.

Discussion:

Diagnostic and management accuracy of SFT is comparable to FTF. Incorrect diagnosis or management of melanomas did not prove to have significant consequences for patient care. Cases subject to analysis of this study are not demographically representative of the general population.

Conclusion:

Diagnosis and management of melanoma in SFT is comparable to FTF care.

Introduction

Store-and-forward teledermatology (SFT) has gained wide acceptance as a means for delivering specialist care. While diagnostic accuracy and management for a variety of conditions have been shown to be non-inferior to face-to-face (FTF) dermatology, concerns remain regarding the safety of SFT in the diagnosis and management of malignant melanoma.^1,2

Part of the concern has centered on the ability of primary care providers to identify and image suspicious lesions. If these lesions are not imaged, they cannot be brought to the attention of the SFT reader. In a recently completed quality improvement study of our teledermatology program, a frequency of 10.1 un-imaged melanomas per 10,000 SFT consultation requests was computed.³ Results from other published studies that address this issue vary widely.^1,4,5

The remainder of the concern has centered on the ability of readers to diagnose and manage the lesion in question. Several publications have addressed the issue of diagnostic accuracy for melanoma in FTF dermatology.^6

–10 In one study, the addition of dermoscopy improved diagnostic accuracy by 49%.¹⁰ A meta-analysis of nine comparable prospective studies confirms this finding.¹¹

Few studies regarding management of malignant melanoma through SFT have been published. However, Warshaw et al. stated that of 36 melanomas in 542 veterans with pigmented lesions, up to 7 (19%) would have been mismanaged in SFT.⁸

These studies in aggregate lead to concern that SFT diagnosis and management of melanomas may be inferior to FTF care. We undertook this quality improvement study to determine the frequency of correct diagnosis and management of primary cutaneous malignant melanoma in a population of veterans who received care in our SFT program. We report a low frequency of incorrectly diagnosed and managed melanomas and discuss contributing factors.

Materials and Methods

Definitions

Melanoma

A primary cutaneous neoplasm diagnosed by pathology with the terms melanoma, melanoma in situ, or lentigo maligna.

Correctly diagnosed melanoma

A melanoma was defined as correctly diagnosed if it met the following criteria: (1) Lesion is contained in at least one image; (2) SFT reader includes “melanoma” or “lentigo maligna” or a similar diagnosis in the differential diagnosis; and (3) The lesion was diagnosed by pathology as melanoma.

Incorrectly diagnosed melanoma

Melanoma meets criteria (1) and (3) but fails to meet criterion (2).

Correctly managed melanoma

A melanoma was defined as correctly managed if it met the following criteria: (1) Lesion is contained in at least one image; (2) Reader includes biopsy in the immediate treatment plan; and (3) Lesion was diagnosed by pathology as melanoma.

Incorrectly managed melanoma

Melanoma meets criteria (1) and (3) but fails to meet criterion (2).

Study Design and Data Sources

This quality improvement study used a retrospective design. This study was deemed “quality improvement” by the Research and Development Service Line, not requiring Institutional Review Board (IRB) review. Data were obtained by query of the Department of Veterans Affairs (VA) Corporate Data Warehouse (CDW), Veterans Information Systems and Technology Architecture (VistA), Veterans Integrated Service Network (VISN) 20 Teledermatology Program database, and by abstraction of the Computerized Patient Record System (CPRS).

Location

Veterans Integrated Service Network 20

VISN 20 encompasses 135 counties in the states of Alaska, California, Idaho, Montana, Oregon and Washington—almost 1,000,000 square miles. Forty-five percent of veterans who receive VA primary care in VISN 20 live in rural or highly rural areas and have little access to VHA dermatology specialist care.

Population

VISN 20 veterans enrolled in SFT.

Inclusion Criteria

• Suspicious lesion imaged by SFT between July 1, 2009 and December 31, 2011.

• Lesion at the site imaged was diagnosed as melanoma by pathology within 1 year of the first SFT consultation.

Exclusion Criteria

• Metastatic and recurrent melanoma.

Data Analysis Procedure

We identified all veterans seen by SFT between July 1, 2009 and December 31, 2011 by VistA query. Possible primary cutaneous malignant melanomas occurring in this cohort of SFT veterans were identified using CDW queries for Systematized Nomenclature of Medicine (SNOMED), International Statistical Classification of Diseases and Related Health Problems (ICD)-9, and Problem List codes. We then conducted a review of the electronic medical record and applied inclusion and exclusion criteria to ascertain all evaluable melanomas occurring in SFT veterans during the period of observation. When the evaluable melanomas were ascertained, we conducted a further review of the images, consultation and pathology reports, applying definitions for correctly diagnosed melanoma and correctly managed melanoma to each case.

Secondary Analysis

To discover potential factors contributing to incorrect diagnosis and management, we ascertained lesion location, patient history, SFT reader identity, and whether dermoscopy was utilized. Images of each lesion were assigned a global quality rating.

Results

During the period of observation, 7,960 unique veterans participated in SFT; 61 melanomas met inclusion and exclusion criteria. Of the 61 melanomas, SFT readers correctly diagnosed 45 (74%) and correctly managed 57 (93%).

Secondary analysis on the subsets of incorrectly diagnosed or incorrectly managed melanomas was performed to ascertain reasons for failure. Since the number of incorrectly managed cases is small, we confine our discussion to the incorrectly diagnosed cases. The likelihood of incorrectly diagnosing a melanoma was not significantly different for in situ versus invasive melanoma, dermatologist clinical experience (≥5 years vs. <5 years), dermatologist SFT experience (≥500 consultations vs. <500 consultations), image quality, or use of dermoscopy (Table 1).

Table 1.

Teledermatology Diagnosis and Management of Melanoma

	CORRECTLY DIAGNOSED	INCORRECTLY DIAGNOSED	CORRECTLY MANAGED	INCORRECTLY MANAGED
All melanomas, n (%)	45 (74)	16 (26)	57 (93)	4 (7)
Pathologic stage, n (%)
pTis	17 (81)	4 (19)	19 (90)	2 (10)
Invasive	28 (70)	12 (30)^a	38 (95)	2 (5)
Reader consult volume (number of cases), n (%)
Reader 1 (6960)	22 (71)	9 (29)	29 (94)	2 (6)
Reader 2 (2191)	12 (80)	3 (20)	15 (100)	0 (0)
Reader 3 (2115)	5 (71)	2 (29)	5 (71)	2 (29)
All others below 500	6 (75)	2 (25)^b	8 (100)	0 (0)
Reader experience, n (%)
<5 years from residency	5 (83)	1 (17)	6 (100)	0 (0)
≥5 years from residency	40 (73)	15 (27)^c	51 (93)	4 (7)
Image quality, n (%)
Low quality	5 (63)	3 (37)	7 (88)	1 (12)
High quality	40 (75)	13 (25)^d	50 (94)	3 (6)
Dermoscopy images present, n (%)
Yes	10 (100)	0 (0)	10 (100)	0 (0)
No	35 (69)	16 (31)^e	47 (92)	4 (8)
Current vital status^f
Living	38	12	46	4
Dead	7	4	11	0
Cause of death melanoma	3	1	3	0
Number of days to biopsy, (mean, median)	(33, 21)	(78, 39)	(44, 22)	(51, 41)

OR 1.82, 95% CI 0.51–6.56, p = 0.36.

OR 0.93, 95% CI 0.17–5.15, p = 0.93.

OR 1.88, 95% CI 0.20–17.39, p = 0.58.

OR 0.54, 95% CI 0.11–2.58, p = 0.77.

OR 9.76, 95% CI 0.54–176.78, p = 0.12.

Current vital status as of October 4, 2016.

Incorrect diagnosis resulted in a difference of median time to biopsy of 18 days. Of the veterans whose melanomas were incorrectly diagnosed, 12 are still alive and clinically free of melanoma (1,769 mean days to follow-up), 4 have died (884 mean days of survival), 1 of melanoma (681 days of survival). Of the veterans whose melanomas were incorrectly managed, all four remain alive (1,718 mean days to follow-up) and are free of melanoma.

Discussion

The principal findings of this quality improvement study are as follows: 45 (74%) melanomas were correctly diagnosed and 57 (93%) were correctly managed.

SFT readers correctly diagnosed 74% of melanomas. This compares favorably to a meta-analysis of diagnostic sensitivity in FTF clinical settings versus dermoscopy in which 71% of FTF cases were correctly diagnosed.¹¹ This value is also comparable to that obtained in a publication by Piccolo et al. where un-blinded dermatologists were e-mailed clinical and dermoscopic images of 43 pigmented lesions. The dermatologists correctly diagnosed 72.7% (8 of 11) of melanomas.¹²

We found that SFT readers correctly diagnosed 68.6% (35/51) and 100% (10/10) of melanomas without and with dermoscopy, respectively. Because of the small number of cases in which dermoscopy images were provided, there is no statistical significance to this observation. Taken with Vestergaard's meta-analysis showing that melanoma diagnosis was 71% sensitive without dermoscopy and 90% with dermoscopy in a FTF clinical setting,¹¹ we believe that the time and resources spent acquiring dermoscopic images adds value to the consultation request and accuracy to the completed report.

Most of our readers had decades of experience as clinical dermatologists. We anticipated that the change to an SFT paradigm for diagnosis might alter diagnostic accuracy for melanomas. Accordingly, our results show that there was remarkable consistency among readers, with the lower volume readers performing as well as the higher volume readers. We conclude that experience as an SFT reader is not a factor in diagnostic accuracy.

There was no statistically significant difference in diagnostic accuracy comparing readers with a few years versus decades of overall FTF clinical experience. We conclude that overall clinical experience is neither an advantage nor disadvantage in diagnostic accuracy in SFT.

Management accuracy of melanoma has been discussed by Warshaw et al., who reported that only 29 of 36 (81%) melanomas would have been correctly managed by SFT readers using macro images only. Adding contact immersion dermoscopy (CID) would have increased this number to 30 (83%).⁸ Incorrect management of melanomas in our study resulted in a delay of 19 days in the median time to biopsy and had no measureable effect on mortality. Despite the delay associated with incorrect management, the safety profile remains uncompromised. Management accuracy in our study, although higher than predicted by previous studies, has yet to be compared to FTF care. A follow-up study is in progress.

Working without CID images, SFT readers appear to perform as well as dermatologists in FTF clinical settings regarding the diagnosis and management of melanoma. Evidence from the literature and preliminary results from our study show enhancement in diagnostic and management accuracy by inclusion of CID images in SFT consultation requests. We are implementing policy changes to require CID images in SFT consultation requests.

Although previous research has investigated diagnostic accuracy of melanoma in FTF settings, no “gold standards” exist for diagnosis and management accuracy.^13,14 We are currently conducting a retrospective cross-sectional study to ascertain diagnostic and management accuracy in FTF dermatology care with the same patient population.

Limitations

If SFT-imaged melanomas were biopsied in the private sector and not reported and documented in the VA electronic medical record, our data would report artificially high rates of correct diagnosis and management. Additionally, melanomas biopsied more than 1 year after imaging were not ascertained; if any occurred, overestimation of the diagnosis and management accuracy would result.

Generalizability of our results may be limited by the fact that this study was conducted in a VHA healthcare system with unique infrastructure defined by standardized protocols, an integrated health records system, and continuing education for teledermatology imagers and primary care providers. These elements may not exist in other teledermatology systems and when absent, have been shown to negatively impact diagnosis and management of skin conditions.¹⁵

Conclusion

Diagnostic accuracy of primary cutaneous melanoma in our SFT program is comparable to FTF dermatology care. Even when melanomas were incorrectly diagnosed or managed, patient outcomes were not significantly affected. This study provides evidence for the safety and effectiveness of SFT as a system for the diagnosis and management of malignant melanoma.

Footnotes

Acknowledgments

This quality improvement study was supported by the Veterans Health Administration. There was no funding to support this effort. This study was deemed “quality improvement” and did not require IRB approval.

Disclosure Statement

No competing financial interests exist.

References

Aldridge

, Naysmith

, Ooi

. The importance of a full clinical examination: Assessment of index lesions referred to a skin cancer clinic without a total body skin examination would miss one in three melanomas. Acta Derm Venereol, 2013; 93:689–692.

Banashur

, Shannon

, Tejasvi

, Kvedar

, et al. The empirical foundations of teledermatology: A review of the research evidence. Telemed J E Health, 2015; 21:953–979.

Gendreau

, Gemelas

, Wang

, et al. Un-imaged melanomas in store and forward teledermatology. 2017. [Epub ahead of print]; DOI: 10.1089/tmj.2016.0170.

Kingsley-Loso

, Grey

, Hanson

, et al. Incidental lesions found in veterans referred to dermatology: The value of a dermatologic examination. J Am Acad Dermatol, 2015; 72:651–655.

Viola

, Tolpinrud

, Gross

, et al. Outcomes of referral to dermatology for suspicious lesions: Implications for teledermatology. Arch Dermatol, 2011; 147:556–560.

Warshaw

, Gravely

, Nelson

, et al. Reliability of store and forward teledermatology for skin neoplasms. J Am Acad Dermatol, 2015; 72:426–435.

Warshaw

, Hillman

, Greer

, et al. Teledermatology for diagnosis and management of skin conditions: A systematic review. J Am Acad Dermatol, 2011; 64:759–772.

Warshaw

, Lederle

, Grill

. Accuracy of teledermatology for pigmented neoplasms. J Am Acad Dermatol, 2009; 61:753–765.

Whited

. Teledermatology research review. Int J Dermatol, 2006; 45:220–229.

10.

Kittler

, et al. Diagnostic accuracy of dermoscopy. Lancet Oncol, 2002; 3:159–165.

11.

Vestergaard

, Macaskill

, Holt

, Menzies

. Dermoscopy compared with naked eye examination for the diagnosis of primary melanoma: A meta-analysis of studies performed in a clinical setting. Br J Dermatol, 2008; 159:669–676.

12.

Piccolo

, Smolle

, Argenziano

, et al. Teledermoscopy-results of a multicentre study on 43 pigmented skin lesions. J Telemed Telecare, 2000;:132–137.

13.

Grin

, Kopf

, Welkovich

, Bart

, Levenstein

. Accuracy in the clinical diagnosis of malignant melanoma. Arch Dermatol, 1990; 126:763–766.

14.

Kopf

, Mintzis

, Bart

. Diagnostic accuracy in malignant melanoma. Arch Dermatol, 1975; 111:1291–1292.

15.

Yen

, et al. Choice, transparency, coordination, and quality among direct-to-consumer telemedicine websites and apps treating skin disease. JAMA Dermatol, 2016; 152:775–68.