mHealth Approaches in Managing Skin Cancer: Systematic Review of Evidence-Based Research Using Integrative Mapping

Background mHealth, which encompasses mobile health technologies and interventions, is rapidly evolving in various medical specialties, and its impact is evident in oncology. In particular, mHealth has established itself as a prominent part of dermatology for cancer screening. Intensified research to seek its use and effectiveness in each phase of the skin cancer continuum is needed in this fast-growing field of teledermatology. Objective The purpose of this review was to describe current trends in research addressing the integration of mHealth and its contributions across the skin cancer continuum. Methods A systematic review framework was applied to the search using three electronic databases: PubMed, Web of Science, and Embase. We extensively reviewed appropriate studies regarding skin cancer and mobile technology published between 2007 and 2017. Studies of the role and impact of mobile technology in the prevention and management of skin cancer were included. We selected 18 studies adhering to the inclusion and exclusion criteria for analysis. Results Of the 18 studies, 5 (28%) evaluated prevention interventions, 6 (33%) assessed diagnostic accuracy, and 7 (39%) pertained to feasibility in the context of mHealth approaches for skin cancer care. These studies portray the potential of mobile teledermatology in the prevention and management of skin cancer. However, not all phases of skin cancer involve mHealth, and not all have been addressed by research. Conclusions This review extends our knowledge not only on the contributions of mHealth technologies, but also on their integration in different phases of skin cancer care. To optimize the effectiveness of mHealth in dermatology, larger numbers of robust, evidence-based studies on teledermatology implementations, distributed evenly across the care continuum, should be conducted so that research can be expanded to systematic reviews.


USA
Mobile phone text messages N=70 Aged 18 years or older n=35 (control group with no text-message reminders) n=35 (intervention group with textmessage reminders) To evaluate the effectiveness of cellular telephone text messaging as a reminder tool for improving adherence to sunscreen application. . RCT; half of the participants received daily text-message reminders via cellular telephone for 6 weeks, and the other half did not receive reminders. The text-message reminders consisted of 2 components: a "hook" text detailing local weather information and a "prompt" text reminding users to apply sunscreen.
Adherence to sunscreen application measured by the number of days participants applied sunscreen during the 6week study period.
The 35 participants who did not receive reminders had a mean daily adherence rate of 30/0% (95% confidence interval, 23.1%-36.9%). The 35 participants who received daily text message reminders had a mean daily adherence rate of 56.1% (95% confidence interval, 48.1%-64.1%). Among the participants in the reminder group, 69% reported that they would keep using the text-message reminders after the study, and 89% reported that they would recommend the text-message reminder system to others. Subgroup analysis did not reveal any significant demographic factors that predicted adherence. Szabo et al., 2015 [15] Hungary Mobile phone text messages N=149 (male=43,female=106) Age: 18 years or older (mean age=36.94 years) n=50 (group 1; no message) n=50 (group2; no message) n=49 (group 3 intervention group) To improve sun protection habits of a sample population using cellular telephone text messaging and e-mails.
Non-blinded, RCT; messages were tailored to individuals, in the form of personalizing with greetings. One group received weekly electronic messages and two groups did not. Intervention group was sent 9 email packages and 3 SMS messages. To test the impact of a theory-based, text message-delivered behavioral intervention (Healthy Text) targeting sun projection or skin selfexamination behaviors to attention control RCT; Each of the three groups received 21 text messages about their assigned topic over 12 months (12 weekly messages for 3 months, then monthly messages for the next 9 months). Messages were personalized using participants' name and gender, skin cancer risk factors, number of times being burnt, previous performance of skin self-examination; telephone surveys sun protection (habits, sunburn, sun tanning), skin self-examination outcomes, sociodemographic outcomes, behavior change, overall level of satisfaction with the intervention, importance of allocated health behaviors, and measures of engagement to the intervention The sun protection (mean change 0.12; P=0.030) and skin self-examination groups (mean change 0.12; P=0.035) significantly greater improvement in their sun protection than the attention control g r o u p . The i n c r e a se i n t h e proportion of participants who reported any skin selfexamination was significantly greater in the skin self-examination group (63%) than the sun protection (48%) or control groups (36%)

USA
Mobile application o n smartphones N=202 (male= female= A volunteer sample of adults aged 18 or order who owned an Android or iPhone smartphones To evaluate a smartphone mobile application (Solar Cell) delivering real-time sun protection advice for a second time in a randomized trial Pre-post RCT; for eight weeks, mobile application gave feedback on sun protection (sun safety practices and sunburn risk) and alerted users to apply/reapply sunscreen and get of the sun; displayed the hourly UV Index and vitamin D production based on the forecast UV Index, time, and location Baseline, three-and eightweek post surveys assessed sun exposure and sun p r o t e c t i o n p r a c t i c e s . Outcome measures included number of days practicing seven sun protection b e h a vi o r s an d days and minutes outdoors in t h e m i d d a y s u n a n d number of sunburns in the past three months.
At 7-weeks, participants assigned to use the app reported wearing widebrimmed hats when outdoors in the midday sun on a greater percentage of days than controls (F=4.72, p=0.03). Males who used the app used less sunscreen lip balm than those no using the app (F=4.19, p=0.04). Women using the app reported using more of all sun protection practices combined than those not using the app. Educated participants, who used the app reported spending fewer days outdoors in the midday sun at the 7week interim survey than controls (F=3.55, p=0.03). To evaluate a smartphone mobile application (Solar Cell) delivering real-time sun protection advice for a second time in a randomized trial Pre-post RCT with a ten-week follow-up; The mobile app provided advice on sun protection (protection practices and risk of sunburn) and alerts (to apply or reapply sunscreen and get out of the sun), hourly UV Index and vitamin D production based on the forecast UV Index, the phone's time and location, and user input.
Outcome measures included the number of days and number of hours spent in the sun, seven sun protection behaviors; sunburn prevalence was assessed with participants' experience of sunburn and the number of sunburns in the past 3 months. Individuals using the app reported a larger mean percentage of time practicing all sun protection behaviors combined than nonusers. Participants not employed reported more days wearing wide-brimmed hats when using the app than those not using it (F=8.57, p<0.01). Using the app was associated with reporting spending fewer hours outdoors in the midday sun by women than men (F=4.88, p=0.03)

Australia
Apple iPhone 4 or 5 with dermascope attachment (FotoFinder Systems) N=230 (all participants who completed preteledermoscopy survey n=49 (among t h e t o t a l , those who conducted mobile TD and completed post-TD survey. Aged 50-64 years at high risk of melanoma (fair skin or previous skin cancer) To assess the self-reported consumer technology acceptance of mobile teledermoscopy when used for early melanoma detection during skin selfexamination Pre-post cohort study; a 27-item questionnaire completed preteledermoscopy.
Postteledermoscopy group, participants conducted a skin self-examination u s i n g m o b i l e teledermoscopy in their homes for early melanoma detection. Submitted their dermoscopic and anatomical images to the study researchers and questionnaire via prepaid mail.

Perceived
usefulness, perceived ease of use, attitude and intention, compatibility, facilitators, subjective norms, and trust. Post-teledermoscopy survey assessed the participants' satisfaction with mobile teledermoscopy, confidence conducting SSE alone or with mobile teledermoscopy and future use.
(Pre-teledermoscopy survey) 48% of items was rated as 'strongly agree' or 'agree' by 75% or more participants. Participants agreed in perceived ease of use, facilitators, subjective norms, attitude/intention and perceived usefulness. 86% agreed mobile TD motivated them to conduct SSE regularly also agreed with the ease of dermatoscope. 78% would use mobile TD again in the future. Barriers: photograph in a 'hard-to-see' body location (18%) and difficulty with photograph submission (35%). Diagnostic accuracy method; one clinical image and one dermoscopic image without clinical data were sent to 2 teleconsultatns, who reviewed the cases independently and answered directly on the web application. Telediagnoses were compared with the face-to-face diagnosis which was taken as correct.
Diagnostic agreement (the concordance between the telediagnosis and the faceto-face diagnosis). Quality of each image (poor, fair, good, excellent) The interobserver agreement among the two teleconsultants was 89% and 94% for the clinical and dermoscopic telediagnoses, respectively. Quality of clinical images judged poor for 31%, fair for 39%, good for 19%, and excellent for 4% of the cases. Dermoscopic images: 11% poor, 42% fair, 42% good, 5% excellent image quality.

Rural areas i n France
Mobile cellular phone (iPhone) teledermoscopy N=289 (male=194, female=95); Aged 18 years or older agricultural population Skin lesions grouped into four management categories: no further treatment/follow-up required, follow-up at 12month intervals, patient advised to make a dermatologist appointment, referral to a local dermatologist for a rapid race-to-face examination. To investigate whether smartphone teledermoscopic referrals could provide faster management of patients with skin cancer and more accurate prioritization of patients with skin lesions of concern compared to normal paper referrals without images One-year case-control, prospective observational study; the general physician takes one clinical and one dermoscopic image with relevant clinical information. Upon sending the referral, the data is embedded on a secure web-based TDS platform.
Simultaneously dermatologists receive an e-mail that a new referral is ready for assessment. The dermatologists log on to the platform to review the referrals on a display monitor and chooses from standardized triage responses Outcome measures the final clinical or histopathological diagnosis o f the referral lesions, the primary therapy applied, the number of visits needed, the waiting times from referral to diagnosis and/or treatment and any incidental findings.
Dermatologist's response in a mean time of 233 min (3.9h). Fastest response was 2 min and the slowest 46h. 79 referrals (98%) assessed in 24h. Paper referrals arrived after a mean time of 5 days (0-82 days). TDS group: requiring surgery with a diagnosis of malignant melanoma shorter waiting time for a first visit with a dermatologist (p<0.0001). Shorter waiting times to receive surgical treatment in the TDS group (P<0.0001). Median waiting time of 36 days for diagnosis and treatment for malignant lesion patients for TDS referrals and 85 days for paper-based referrals (p<0.0001). Decisions on the referral of patients with suspicious skin lesions by store-andforward TD vs. by a conventional referral system. Breslow thickness and tumor stage were recorded in each study group (TD and non-TD) and were compared.
The frequency of melanoma with a favorable initial prognosis was significantly higher in the TD group (70.1% vs 56.9%, P=0.03). The odds ratio of having a cutaneous melanoma with a favorable initial prognosis in the TD group was 1.96 (95% CI, 1.14-3.50; P=0.04)

Egypt
Sotware-enabled mobile telephone containing a 5megapixel camera

N=30
To demonstrate that mobile telephones may be used on the African continent to submit both patient history and clinical photographs wirelessly to remote expert dermatologists, and to assess whether these data are diagnostically reliable.
The diagnoses formulated through face-to-face examination were tallied and compared with the diagnoses formulated through teleconsultation and diagnostic concordance rates were tabulated.
Feasibility was assessed based on diagnostic agreement between face-toface consultation and local dermatologists' independent evaluation by teleconsultation Diagnostic agreement between faceto-face consultation and the two local senior dermatologist performing independent evaluation by teleconsultation was achieved in 23/30 (70%) and in 22/30 (73%) cases, respectively (global mean 75%).
Mobile teledermatology is regarded as a technically feasible and diagnostically reliable method of amplifying access to dermatologic expertise. Over a 3-month period up to three clinical and dermoscopic images were obtained of 113 skin tumors from 88 patients using a mobile phone camera. Clinical and dermoscopic images of each lesion together with clinical information were separately transmitted for decision-making. The lesions were grouped into four diagnostic categories, and diagnoses were then compared with the gold standard.
Diagnostic concordance (results were compared with those obtained by faceto-face examination and histopathology as the gold standard).
Clinical and dermoscopic teleevaluations demonstrated strong concordance with the gold standard (k=0.84 for each) and similar high sensitivity and specificity for all diagnostic categories. Clinical image tele-evaluation was superior to teledermoscopy. To estimate the diagnostic performance (sensitivity, specificity, and predictive values) of lesion-directed mobile teledermatology with or without dermoscopic imaging as a skin cancer screening intervention in a population representing a low-prevalence setting Prospective diagnostic study; Images arranged by skin lesions were organized in 2 packages, the first one macroscopic images only and the second one with additional teledermascopic images, forwarded to the teledermatologist. TD first decided adequacy of images to allow assessment for cancer screening.
In a d e q u a t e c a s e s , recommendations were given in 3 categories. a dermatologist meeting the patient faceto-face; to assess the adequacy of the TD's management decisions; to evaluate the image quality obtained During a 16-week period, patients with one or more suspicious skin lesions in need of a biopsy or excision included; the smartphone app used to send a clinical image (dermoscopy image and clinical information) to Internet platform. Two TDs assessed the incoming cases, providing a specific primary diagnosis and a management decision.
Histopathological diagnosis was used as the gold standard.
Diagnostic accuracy of the face-to-face (FTF) dermatologist and TDs; interobserver concordances; image quality; level of diagnostic difficulty. 69 lesions included; Diagnostic accuracy: 66.7% (FTF), 50.7% (TD 1), 60.9% (TD 2). Moderate to substantial agreement between TFT dermatologist and the TDs and between the TDs for interobserver concordances. The TDs provided adequate management decisions for 98.6% and 100% lesions, respectively. Image quality rated as excellent or sufficient in 94% and 84% of the cases by the respective TDs. To determine the diagnostic concordance between the conventional face-to-face diagnosis and the telediagnosis of 10 dermatologists with expertise in dermatooncology of 10 challengeing pigmented lesions.
Clinical and dermoscopic digital images of all selected lesions transmitted via e-mail to 10 dermatologists using a storeand-forward teledermatology method. Dermatologists provided teled iagno ses with a step-bystep approach. Diagnostic concordance; final histopathological diagnosis was considered the gold standard for comparison with a face-toface and teledermatology diagnoses in statistical analysis.
Face-to-face results indicated moderate agreement between clinical and histopathological diagnoses (k=0.6); interobserver concordance of telediagnosis lower than face-to-face diagnosis (k=0.52); the concordance declined f u r t h e r (k=0.38) after the second dermoscopy step.