Teledermatology consultation using a smartphone multimedia messaging service for common skin diseases in the Korean army: a clinical evaluation of its diagnostic accuracy

Abstract

Summary

We evaluated the diagnostic accuracy of teleconsultations for skin diseases common in the army using a smartphone multimedia messaging service (MMS). Images of skin lesions were obtained from 100 army patients using digital cameras built into smartphones. Three remotely located dermatologists received the dermatology images and associated clinical information via the MMS. The teledermatologists’ diagnoses were compared with those obtained from face-to-face examinations. The three most common diagnoses made at the dermatology clinics were eczema, viral warts and fungal infections. The mean diagnostic agreement between face-to-face and teledermatology consultations was 71% (SD 2). The mean kappa coefficient was 0.73 (SD 0.06) for the three most common diagnostic categories. The mean values for sensitivity were 78% (SD 0), 88% (SD 21) and 61% (SD 11) for eczema, viral warts and fungal infections, respectively, and the specificity values were above 90% for these skin diseases. Teledermatology consultation using smartphones is simple. Although diagnoses using telemedicine do not perfectly match diagnoses from face-to-face consultations the diagnostic accuracy using smartphones is superior to that of clinicians who are not specialized in dermatology.

Introduction

Although people living in urban areas of Korea usually have ready access to dermatology clinics and hospitals, people in rural areas or in the army may not. This is a particular problem for people in the army, because there are few medical officers who specialize in dermatology, and dermatology clinics are usually located far from the military units.

Military service is mandatory for all young and healthy Korean men. The prevalence of skin diseases in the Korean army is very high at 60%, and some of the conditions can significantly affect daily life.¹ Since there are not enough dermatologists in the army to treat these skin conditions, a teledermatology consultation system would allow primary caregivers to consult with dermatologists from a distance. In the US Army, dermatology consultations comprise 31% of its telemedicine consultation service.²

Store and forward (asynchronous) telemedicine can be undertaken using mobile phones because they usually contain built-in digital cameras and the multimedia messaging services (MMS) available to smartphones can be used to transfer the images for teledermatology consultations without the need for any other technology.

While previous studies have shown that teledermatology diagnoses are fairly accurate compared with conventional face-to-face consultations,^3–6 no studies, to our knowledge, have evaluated the diagnostic accuracy of teledermatology consultations carried out with army personnel using smartphones and MMS. The aim of the present study was to investigate the diagnostic accuracy of teledermatology consultations for common skin diseases among army personnel using a smartphone and MMS.

Methods

Patients were enrolled prospectively from dermatology clinics held at the Armed Forces Yangju Hospital. Informed written consent was obtained from all patients who enrolled in the study. The inclusion criteria were that the participants were engaging in their military service, they visited the dermatology clinics with visual skin lesions, and they agreed to participate in this study. The exclusion criteria were a complaint without visual skin lesions, or a refusal to participate in the study. The study protocol was approved by the appropriate ethics committees.

Tele-evaluation

A paramedic with no specialist dermatology knowledge or experience took digital photographs of the skin lesions using the camera built into a smartphone (IM-A830L, Pantech, Seoul, Korea). This produced images of 8 Mpixel resolution. The paramedic was not specially trained to take pictures of skin lesions, i.e. the situation was intended to simulate real life. The digital photographs and a brief text message containing information about the patient’s age, the duration and symptoms of the lesion, any signiﬁcant medical history, and/or any concomitant medication were sent using the MMS (Kakaotalk, Kakao, Sungnam, Korea) (Figure 1). Three dermatologists who worked at the Seoul National University Hospital about 40 km from Yangju, received these messages on their own smartphones (Teledermatologist 1: iPhone3, Apple, Cupertino, CA, USA; Teledermatologist 2: Galaxy S, Samsung Electronics, Seoul, Korea; Teledermatologist 3: Galaxy S II, Samsung Electronics, Seoul, Korea). The dermatologists had previous experience in teledermatology, both store and forward and real-time. After magnifying the images on their smartphone's display, the three dermatologists then independently decided on the clinical diagnoses.

Figure 1.

Digital photographs and a concise text message were transferred to three dermatologists working remotely, using smartphones and the multimedia messaging service.

After the paramedic had transmitted the images and the patients’ information to the remotely located dermatologists, the patients were sent to the dermatology clinic to meet a specialist dermatologist for face-to-face consultations. The in-person dermatologist was the same person throughout the study. Since not all patients had their skin biopsied, the diagnoses made from the face-to-face consultations were considered to be the gold standard, regardless of any histopathology confirmation.

After the primary diagnoses had been made by the teledermatologists, they were sorted into the diagnostic categories shown in Table 1. Diagnoses were categorized by taking into consideration the clinical characteristics of the skin diseases with reference to a dermatology textbook.⁷ The primary diagnoses given by the teledermatologists were compared with the diagnoses given during the face-to-face consultations. In addition, the primary diagnoses given by the three teledermatologists were compared with each other to test the reliability of the teledermatology consultations.

Table 1.

Diagnostic categories used.

Diagnostic category
1	Eczema
2	Papular squamous disorders
3	Urticaria and erythema
4	Bacterial infections
5	Viral warts
6	Viral disease (excluding warts)
7	Fungal infections
8	Bites and stings
9	Sexually transmitted diseases
10	Benign skin tumours
11	Malignant skin tumours
12	Skin manifestations of systemic disorders
13	Skin injuries
14	Drug eruptions
15	Bullous diseases
16	Acne and folliculitis
17	Disorders of the sweat glands
18	Alopecia areata
19	Androgenetic alopecia
20	Other hair disorders
21	Primary nail disorders
22	Hypermelanocytic disorders
23	Hypomelanocytic disorders
24	Other pigmentary disorders
25	Scars and keloids
26	Connective tissue skin diseases
27	Hereditary skin diseases
28	Factitial skin diseases
29	Miscellaneous skin diseases

Statistical analysis

The frequencies of the skin diseases were counted, and the rates of diagnostic agreement between the face-to-face and the teledermatology consultations were calculated as percentages. Diagnostic agreement was analysed using Cohen’s kappa coefficient, which corrects for the proportion of agreement expected by chance. A kappa coefficient <0.40 indicates fair agreement, a kappa coefficient of 0.41–0.60 indicates moderate agreement, a kappa coefficient of 0.60–0.80 indicates substantial agreement and a kappa coefficient >0.80 indicates almost perfect agreement.⁸ In addition, the sensitivity and specificity of the teledermatology consultations were calculated. Statistical analyses were performed using a standard package (SPSS 21.0, IBM, New York, USA).

Results

One hundred male patients with a mean age of 20.6 years (range 18–26) were enrolled in the study between November 2012 and December 2012. The most common diagnoses based on face-to-face examinations at the clinics were eczema (23%), viral warts (14%) and fungal infections (11%); together these three diagnostic categories comprised 48% of all of the cases (Table 2). There was a similar pattern in the diagnoses made by the teledermatologists (data not shown).

Table 2.

Frequencies of skin conditions diagnosed at dermatology clinics held at the Korean army hospital.

Number	Percent
Eczema	23	23
Viral wart	14	14
Fungal infection	11	11
Skin injury	8	8
Acne and folliculitis	8	8
Bacterial infection	7	7
Benign skin tumour	7	7
Scar and keloid	5	5
Urticaria and erythema	4	4
Hypomelanocytic disorder	3	3
Alopecia areata	2	2
Hypermelanocytic disorder	2	2
Miscellaneous skin disease	2	2
Papular squamous disorders	1	1
Androgenetic alopecia	1	1
Other pigmentory disorder	1	1
Factitial skin disease	1	1
Total	100	100

Diagnostic agreement

The mean agreement between the primary diagnoses made during face-to-face and teledermatology consultations was 70.7% (SD 1.5). The mean kappa coefficient was 0.73 (SD 0.06) for the three most common diagnostic categories, indicating substantial agreement (kappa = 0.70, 0.69 and 0.80 for teledermatologists 1, 2 and 3, respectively). These kappa coefficients were significant (P < 0.05).

Reliability

To test the reliability of the teledermatology consultations, the primary diagnoses made by the three teledermatologists were compared in pairs (Table 3). The mean overall agreement in diagnoses that included all of the diagnostic categories was 83% (SD 7). The mean kappa coefficient was calculated for the three most common diagnostic categories and was 0.70 (SD 0.07). These kappa coefficients were significant (P < 0.05).

Table 3.

Diagnostic agreement between teledermatologists.

Teledermatologists	Agreement (%)	Kappa coefficient	P-value
1 and 2	77.0	0.623	<0.001
1 and 3	82.0	0.712	<0.001
2 and 3	91.0	0.768	<0.001

Sensitivity and specificity

The sensitivity and specificity were calculated for the diagnoses of eczema, viral warts and fungal infections, the three most common diagnostic categories. The mean values for sensitivity were 78% (SD 0), 88% (SD 21) and 61% (SD 11) for eczema, viral warts and fungal infections, respectively. The mean values for specificity were 93.1% (SD 5.2), 99.6% (SD 0.7) and 98.1% (SD 1.7) for eczema, viral warts and fungal infections, respectively.

Discussion

A variety of reasons, including exposure to sunlight, temperature extremes, living uncomfortably in groups, poor sanitation and skin irritation caused by the uniforms, guns, helmets and boots, underlie the high prevalence of skin diseases in the army.² The diagnosis of skin disease is not straightforward for primary caregivers, and diagnoses made by dermatologists are twice as accurate as those made by non-dermatologists.⁹,¹⁰ Consequently, many soldiers are referred to dermatologists who are located a long way from their bases to have their skin conditions diagnosed, even though the complaints may not be serious. Since this causes poor morale and combat ineffectiveness,² a variety of ways to reduce unnecessary evacuations from military units have been considered. The US Army started teledermatology consultations in about 1992, and a formal teledermatology consultation system using email was established in 2004.² A total of 2157 patients used the teledermatology consultation system in Iraq and it saved US$30.4 million.¹¹ This shows the value of teledermatology consultation in the context of the army, because it reduces the number of soldiers who must be withdrawn from frontline duties and it saves money.

However, unlike the US Army the Korean army does not have a formal teledermatology consultation system. Therefore, it was important to investigate the efficacy of teledermatology consultation in the Korean army. Our study indicates that eczema, viral warts and fungal infections, in particular, are very common among Korean soldiers. Indeed, a previous study that investigated skin diseases among military personnel also reported that fungal infections and eczema were the most common skin complaints.¹² If the Korean military authorities can manage the three most common skin diseases, the burden associated with skin diseases in the Korean army will decrease dramatically.

Since the number of dermatologists in the army is limited, it has been suggested that a teledermatology consultation system should be established to manage skin diseases. However, two problems must be overcome. The first relates to the way the system will be financed and the second relates to the clinical accuracy of the system.

To develop and implement a national system for teledermatology consultations would require substantial investment. It would also take a considerable length of time. Hence, in our study we proposed the use of smartphones. The smartphone user penetration rate in Korea is 68%, the highest in the world in 2012.¹³ All of the smartphones used in our study were equipped with high-resolution digital cameras (8 Mpixel), except one that had a 4 Mpixel digital camera. The images and the text were rapidly transferred using the MMS and wireless Internet or third-generation (3G) telephone networks. The whole of South Korea is covered by 3G telephone networks. The MMS, “Kakaotalk”, is widely distributed across Korea and it can transfer high-resolution images to multiple people simultaneously, free of charge. Hence, smartphones in combination with a pre-existing high-speed communication network, allow store and forward teledermatology consultations to be performed with no associated communication costs.

The accuracy of teledermatology consultations has been the focus of several previous studies. A review article about teledermatology demonstrated that the diagnostic accuracy of store and forward teledermatology consultations was inferior to that achieved during face-to-face consultations in dermatology clinics.⁴ Agreement in relation to the primary diagnoses ranged from 46–88% in 15 previous studies that investigated store and forward teledermatology consultations, and the weighted average was 67%.¹⁴ In our study, the rate of concordance between primary diagnoses made in the dermatology clinics and those made by teledermatology was 71%, which is consistent with previous reports. Furthermore, the kappa coefficients determined in our study were in the range 0.65–0.93 that had been determined in previous studies on store and forward teledermatology, even though the number of participants in the study was limited.¹⁴

Unlike previous studies, our investigation was undertaken in a military setting using smartphones. We did not intend everyone to use the same model of smartphone, because we thought using different smartphone models would better represent the conditions encountered in the field.

The rate of diagnostic concordance in our study was similar to those reported from previous studies, and this suggests that smartphones could be readily incorporated into a store and forward system. However, some factors need to be considered before undertaking teledermatology consultations using smartphones. For example, the diagnostic agreement among the teledermatologists in the current study was not very high, with the kappa coefficients ranging from 0.62–0.77, and all were less than 0.80. A high level of reliability is necessary for a diagnostic tool. Moreover, the kappa coefficients determined in our study only reflected about half of our study population. While all of the diagnoses were included when we calculated the rates of concordance, when we computed the kappa coefficients we only included the three most commonly diagnosed diseases. This was because 18 diagnostic categories represented too many variables to calculate the kappa coefficient reliably.

The face-to-face diagnoses did not represent final, confirmed diagnoses because we did not perform histopathology, which is generally regarded as the gold standard for dermatology diagnosis. However, many types of eczema, viral warts and fungal infections do not require biopsies to reach a diagnosis. In the present study, 9% of the patients had skin tumours or hypermelanocytic disorders and in these cases, histopathology examination is usually recommended to diagnose these diseases correctly. Teledermatology consultations could triage such patients to ensure they have biopsies taken.

The sensitivity in our study was lower than that reported from previous studies, which ranged from 0.88--1.0.¹⁴ We only determined sensitivity values for the three most common diagnostic categories. The sensitivity for the diagnosis of fungal infections was only 61%, which suggests that the use of smartphones for teledermatology consultations could miss fungal infections. The mean specificity values for eczema, viral warts and fungal infections was 93.1-99.6%, and we think that misdiagnoses may have been caused by the lack of information provided for the teledermatology consultations. The MMS text messages only contained very brief information about the patients, and the teledermatologists were unable to ask the patients for more information about their medical histories or other symptoms they might have been experiencing

In conclusion, the use of smartphones for teledermatology consultations involves simple processes and has low costs associated with it. In addition, the diagnostic concordance with face-to-face consultations in dermatology clinics is substantial and may be sufficient for routine use in the field. In fact the diagnostic accuracy using smartphones is superior to that of clinicians who are not specialized in dermatology. A valuable future study would involve testing and comparing the accuracy of diagnoses made via teledermatology with those made by clinicians who are not specialized in dermatology.

Footnotes

Acknowledgements

We are grateful for the cooperation of the paramedics at the Armed Forces Yangju Hospital.

References

Bae

Lee

Park

Kim

Park

. Prevalence of common skin diseases and their associated factors among military personnel in Korea: a cross-sectional study. J Korean Med Sci 2012; 27: 1248–54.

McGraw

Norton

. Military aeromedical evacuations from central and southwest Asia for ill-defined dermatologic diseases. Arch Dermatol 2009; 145: 165–70.

Loane

Corbett

Bloomer

. Diagnostic accuracy and clinical management by realtime teledermatology. Results from the Northern Ireland arms of the UK Multicentre Teledermatology Trial. J Telemed Telecare 1998; 4: 95–100.

Whited

Hall

Simel

. Reliability and accuracy of dermatologists’ clinic-based and digital image consultations. J Am Acad Dermatol 1999; 41: 693–702.

Lowitt

Kessler

Kauffman

Hooper

Siegel

Burnett

. Teledermatology and in-person examinations: a comparison of patient and physician perceptions and diagnostic agreement. Arch Dermatol 1998; 134: 471–6.

Phillips

Burke

Shechter

Stone

Balch

Gustke

. Reliability of dermatology teleconsultations with the use of teleconferencing technology. J Am Acad Dermatol 1997; 37: 398–402.

Wolff

Goldsmith

Katz

Gilchrest

Paller

Leffell

. Fitzpatrick's Dermatology in General Medicine, 7th ed. New York, NY: McGraw-Hill, 2008.

Kroemer

Frühauf

Campbell

. Mobile teledermatology for skin tumour screening: diagnostic accuracy of clinical and dermoscopic image tele-evaluation using cellular phones. Br J Dermatol 2011; 164: 973–9.

Ramsay

Fox

. The ability of primary care physicians to recognize the common dermatoses. Arch Dermatol 1981; 117: 620–2.

10.

Tran

Chen

Lim

Jabbour

Shumack

. Assessing diagnostic skill in dermatology: a comparison between general practitioners and dermatologists. Australas J Dermatol 2005; 46: 230–4.

11.

Henning

Wohltmann

Hivnor

. Teledermatology from a combat zone. Arch Dermatol 2010; 146: 676–7.

12.

Selvaag

. Skin disease in military personnel. Mil Med 2000; 165: 193–4.

13.

Bicheno S. Global Smartphone User Penetration Forecast by 88 Countries: 2007 to 2017. See http://www.strategyanalytics.com/default.aspx?mod=reportabstractviewer&a0=8044 (last checked 2 December 2013).

14.

Warshaw

Hillman

Greer

. Teledermatology for diagnosis and management of skin conditions: a systematic review. J Am Acad Dermatol 2011; 64: 759–72.