Evaluation of a skin self examination attitude scale using an item response theory model approach

Web-Based Skin Cancer Assessment and Classification Using Machine Learning and Mobile Computerized Adaptive Testing in a Rasch Model: Development Study

Background

Web-based computerized adaptive testing (CAT) implementation of the skin cancer (SC) risk scale could substantially reduce participant burden without compromising measurement precision. However, the CAT of SC classification has not been reported in academics thus far.

Objective

We aim to build a CAT-based model using machine learning to develop an app for automatic classification of SC to help patients assess the risk at an early stage.

Methods

We extracted data from a population-based Australian cohort study of SC risk (N=43,794) using the Rasch simulation scheme. All 30 feature items were calibrated using the Rasch partial credit model. A total of 1000 cases following a normal distribution (mean 0, SD 1) based on the item and threshold difficulties were simulated using three techniques of machine learning—naïve Bayes, k-nearest neighbors, and logistic regression—to compare the model accuracy in training and testing data sets with a proportion of 70:30, where the former was used to predict the latter. We calculated the sensitivity, specificity, receiver operating characteristic curve (area under the curve [AUC]), and CIs along with the accuracy and precision across the proposed models for comparison. An app that classifies the SC risk of the respondent was developed.

Results

We observed that the 30-item k-nearest neighbors model yielded higher AUC values of 99% and 91% for the 700 training and 300 testing cases, respectively, than its 2 counterparts using the hold-out validation but had lower AUC values of 85% (95% CI 83%-87%) in the k-fold cross-validation and that an app that predicts SC classification for patients was successfully developed and demonstrated in this study.

Conclusions

The 30-item SC prediction model, combined with the Rasch web-based CAT, is recommended for classifying SC in patients. An app we developed to help patients self-assess SC risk at an early stage is required for application in the future.

The Melanoma Genomics Managing Your Risk Study randomised controlled trial: statistical analysis plan

Background

The Melanoma Genomics Managing Your Risk Study is a randomised controlled trial that aims to evaluate the efficacy of providing information on personal genomic risk of melanoma in reducing ultraviolet radiation (UV) exposure, stratified by traditional risk group (low or high phenotypic risk) in the general population. The primary outcome is objectively measured total daily Standard Erythemal Doses at 12 months. Secondary outcomes include UV exposure at specific time periods, self-reported sun protection and skin-examination behaviours, psychosocial outcomes, and ethical considerations surrounding offering genomic testing at a population level. A within-trial and modelled economic evaluation will be undertaken from an Australian health system perspective to assess the cost-effectiveness of the intervention.

Objective

To publish the pre-determined statistical analysis plan (SAP) before database lock and the start of analysis.

Methods

This SAP describes the data synthesis, analysis principles and statistical procedures for analysing the outcomes from this trial. The SAP was approved after closure of recruitment and before completion of patient follow-up. It outlines the planned primary analyses and a range of subgroup and sensitivity analyses. Health economic outcomes are not included in this plan but will be analysed separately. The SAP will be adhered to for the final data analysis of this trial to avoid potential analysis bias that may arise from knowledge of the outcome data.

Results

This SAP is consistent with best practice and should enable transparent reporting.

Conclusion

This SAP has been developed for the Melanoma Genomics Managing Your Risk Study and will be followed to ensure high-quality standards of internal validity and to minimise analysis bias.

Trial registration

Prospectively registered with the Australian New Zealand Clinical Trials Registry, ID: ACTR N12617000691347. Registered on 15 May 2017.

Accuracy of mobile digital teledermoscopy for skin self-examinations in adults at high risk of skin cancer: an open-label, randomised controlled trial

BACKGROUND

Skin self-examinations supplemented with mobile teledermoscopy might improve early detection of skin cancers compared with naked-eye skin self-examinations. We aimed to assess whether mobile teledermoscopy-enhanced skin self-examination can improve sensitivity and specificity of self-detection of skin cancers when compared with naked-eye skin self-examination.

METHODS

This randomised, controlled trial was done in Brisbane (QLD, Australia). Eligible participants (aged ≥18 years) had at least two skin cancer risk factors as self-reported in the eligibility survey and had to own or have access to an iPhone compatible with a dermatoscope attachment (iPhone versions 5-8). Participants were randomly assigned (1:1), via a computer-generated randomisation procedure, to the intervention group (mobile dermoscopy-enhanced self-skin examination) or the control group (naked-eye skin self-examination). Control group and intervention group participants received web-based instructions on how to complete a whole body skin self-examination. All participants completed skin examinations at baseline, 1 month, and 2 months; intervention group participants submitted photographs of suspicious lesions to a dermatologist for telediagnosis after each skin examination and control group participants noted lesions on a body chart that was sent to the research team after each skin examination. All participants had an in-person whole-body clinical skin examination within 3 months of their last skin self-examination. Primary outcomes were sensitivity and specificity of skin self-examination, patient selection of clinically atypical lesions suspicious for melanoma or keratinocyte skin cancers (body sites examined, number of lesions photographed, types of lesions, and lesions missed), and diagnostic concordance of telediagnosis versus in-person whole-body clinical skin examination diagnosis. All primary outcomes were analysed in the modified intention-to-treat population, which included all patients who had a clinical skin examination within 3 months of their last skin self-examination. This trial was registered with the Australian and New Zealand Clinical Trials Registry, ACTRN12616000989448.

FINDINGS

Between March 6, 2017, and June 7, 2018, 234 participants consented to enrol in the study, of whom 116 (50%) were assigned to the intervention group and 118 (50%) were assigned to the control group. 199 participants (98 participants in the intervention group and 101 participants in the control group) attended the clinical skin examination and thus were eligible for analyses. Participants in the intervention group submitted 615 lesions (median 6·0 per person; range 1-24) for telediagnosis and participants in the control group identified and recorded 673 lesions (median 6·0 per person; range 1-16). At the lesion level, sensitivity for lesions clinically suspicious for skin cancer was 75% (95% CI 63-84) in the intervention group and 88% (95% CI 80-91) in the control group (p=0·04). Specificity was 87% (95% CI 85-90) in the intervention group and 89% (95% CI 87-91) in the control group (p=0·42). At the individual level, the intervention group had a sensitivity of 87% (95% CI 76-99) compared with 97% (95% CI 91-100) in the control group (p=0·26), and a specificity of 95% (95% CI 90-100) compared with 96% (95% CI 91-100) in the control group. The overall diagnostic concordance between the telediagnosis and in-person clinical skin examination was 88%.

INTERPRETATION

The use of mobile teledermoscopy did not increase sensitivity for the detection of skin cancers compared with naked-eye skin self-examination; thus, further evidence is necessary for inclusion of skin self-examination technology for public health benefit.

FUNDING

National Health and Medical Research Council (Australia).

Development and initial validation of the Self-Efficacy for Skin Self-Examination Scale in a Canadian sample of patients with melanoma

PURPOSE

Melanoma is the deadliest type of skin cancer, although survival rates are high if detected early. Skin self-examination (SSE) is a health behaviour that can lead to early detection of melanoma and more positive health outcomes. Self-efficacy for SSE is a potential predictor of engaging in skin self-exams. However, no standardized measures of self-efficacy for SSE are currently available. The present study reports on the development and initial validation of a measure assessing self-efficacy for SSE.

METHOD

Based on a literature review, including previous studies assessing this construct, the research team developed 9 items capturing confidence in one's ability to conduct SSE. Items were subsequently revised by the research team and two dermatologists, resulting in the elimination of 4 items and modifications to the response options. The final 5-item Self-Efficacy for SSE scale was administered to a sample of 242 melanoma patients recruited from local hospitals. We assessed the scale's factor structure through exploratory factor analysis and the internal consistency with Cronbach's alpha. Scale scores were correlated with intentions to perform SSE, physician support for SSE, skin cancer-specific distress, general distress, and sociodemographic variables to assess convergent and divergent validity.

RESULTS

The scale was found to be unifactorial with adequate model fit, have acceptable internal consistency (Cronbach alpha = 0.74), and initial evidence of convergent and divergent validity, as indicated by the scale correlating with physicians' support for SSE (r = 0.30, p < 0.01), intention to perform SSE (r_s = 0.21, p < 0.01) and a weak correlation with general distress (r = 0.15, p < 0.05).

CONCLUSIONS

Pending future research corroborating these findings, this measure has potential for use by nurses and other health professionals in follow-up care to identify individuals at risk for low self-efficacy for SSE requiring more targeted educational SSE interventions.

Redesigning Skin Cancer Early Detection and Care Using a New Mobile Health Application: Protocol of the SKIN Research Project, a Randomised Controlled Trial

Patients often detect melanoma themselves; therefore, regular skin self-examinations (SSEs) play an important role in the early detection and prompt treatment of melanoma. Mobile teledermoscopy is a technology that may facilitate consumer SSEs and rapid communication with a dermatologist. This paper describes the planned randomised controlled trial of an intervention to determine whether mobile technologies can help improve the precision of SSE in consumers. A randomised controlled trial will be conducted to evaluate mobile teledermoscopy-enhanced SSE versus naked-eye SSE. Participants in each group will conduct three home whole-body SSEs at baseline, 1 and 2 months, then present for a clinical skin examination (CSE) by a doctor after the 2-month SSE. Specifically, participants will identify skin lesions that meet the AC (asymmetry and colour) rule for detecting a suspicious skin spot. The primary outcomes are sensitivity and specificity of the skin lesions selected by the participants as needing attention by a doctor, compared to the clinical diagnosis by the dermatologist that will serve as the reference standard for this analysis. For the mobile teledermoscopy-enhanced SSE group, researchers will assess the number, location and type of lesions (1) sent by the participant via mobile teledermoscopy, (2) found at CSE or (3) missed by the participant. For the naked-eye SSE group, researchers will assess the number, location and type of lesions (1) recorded on their body chart by the participant, (2) found at CSE or (3) missed by the participant. Secondary outcomes are based on participants' self-reported data via online questionnaires.

The melanoma genomics managing your risk study: A protocol for a randomized controlled trial evaluating the impact of personal genomic risk information on skin cancer prevention behaviors

BACKGROUND

Reducing ultraviolet radiation (UV) exposure and improving early detection may reduce melanoma incidence, mortality and health system costs. This study aims to evaluate the efficacy and cost-effectiveness of providing information on personal genomic risk of melanoma in reducing UV exposure at 12 months, according to low and high traditional risk.

METHODS

In this randomized controlled trial, participants (target sample = 892) will be recruited from the general population, and randomized (1:1 ratio, intervention versus control). Intervention arm participants provide a saliva sample, receive personalized melanoma genomic risk information, a genetic counselor phone call, and an educational booklet on melanoma prevention. Control arm participants receive only the educational booklet. Eligible participants are aged 18-69 years, have European ancestry and no personal history of melanoma. All participants will complete a questionnaire and wear a UV dosimeter to objectively measure their sun exposure at baseline, 1- and 12-month time-points, except 1-month UV dosimetry will be limited to ~250 participants. The primary outcome is total daily Standard Erythemal Doses at 12 months. Secondary outcomes include objectively measured UV exposure for specific time periods (e.g. midday hours), self-reported sun protection and skin-examination behaviors, psycho-social outcomes, and ethical considerations surrounding offering genomic testing at a population level. A within-trial and modelled economic evaluation will be undertaken from an Australian health system perspective to assess the intervention costs and outcomes.

DISCUSSION

This trial will inform the clinical and personal utility of introducing genomic testing into the health system for melanoma prevention and early detection at a population-level.

TRIAL REGISTRATION

Australian New Zealand Clinical Trials Registry ACTRN12617000691347.

Impact of hypoglycemic episodes on health-related quality of life of type-2 diabetes mellitus patients: development and validation of a specific QoLHYPO© questionnaire

Background

Hypoglycemia is a limiting factor to achieving optimal glycemic control in patients with type-2 diabetes mellitus (T2DM), increasing risk of death and complications, reducing health-related quality of life (HRQoL) and work productivity and increasing healthcare costs.

The study’s primary objective was to develop and validate a specific questionnaire to assess the impact of hypoglycemia on the HRQoL of T2DM patients (QoLHYPO^© questionnaire).

Methods

A two-phase multicenter prospective, longitudinal, observational, epidemiologic study of consecutively enrolled patients, not involving any drug, was conducted: In phase 1 (questionnaire development), patients who had given their written informed consent, who were at least 30 years of age, had been diagnosed with T2DM at least 5 years prior, had an HbA1c test in the previous 3 months, and a hypoglycemic episode in the previous 6 months were included. To validate the questionnaire and assess reliability and responsiveness, phase 2 included two cohorts of patients. Patients in the reliability cohort would likely have stable clinical course during the 3 weeks following inclusion in the study and patients in the responsiveness cohort would likely experience changes in their clinical course in the 3 months after enrollment.

Results

Phase 1 included 168 patients: 10 attended semi-structured interviews, 18 for face validity, and 140 for the pilot test (Rasch analysis). Phase 2 included 227 patients: 142 in the reliability cohort and 85 in the responsiveness cohort.

Of the 37 items initially included in Phase 1, 11 (floor/ceiling effect analysis) and 13 (Rasch analysis) were discarded. The final version of the questionnaire consisted of 13 items.

Phase 2 results showed the questionnaire was unidimensional and able to accurately assess HRQoL. Intra-observer reproducibility (ICC = 0.920) and internal consistency (Cronbach’s alpha: visit 1 = 0.912; visit 2 = 0.901) were high, showing high reliability. Internal responsiveness was moderate (standardized effect size 0.5-0.8) and external responsiveness was lower (AUC > 0.5; not statistically significant). Minimal clinically important difference (MCID) was estimated to be 3.2 points.

Conclusions

The QoLHYPO^© questionnaire is a tool that can be used in routine clinical practice to assess the impact of hypoglycemia on the HRQoL of T2DM patients.

Electronic supplementary material

The online version of this article (10.1186/s12955-018-0875-1) contains supplementary material, which is available to authorized users.

Rasch analysis in the development of the NutriQoL® questionnaire, a specific health-related quality of life instrument for home enteral nutrition

Background

Home enteral nutrition (HEN) is a therapeutic method used in patients who are unable to ingest the required amounts of nutrients but retain a functional gastrointestinal tract. The objective of this study was to compose a specific questionnaire for measuring health-related quality of life (HRQoL) in HEN patients irrespective of their underlying condition and HEN route of administration.

Methods

Literature review, focus groups and semi-structured interviews were used to propose an initial version of the questionnaire which was answered by 165 participants. The responses were analyzed using the Rasch methodology. Firstly, the appropriateness of response options was assessed. Then, the differential item functioning (DIF) was evaluated. Finally, the item fit statistics, infit and outfit, were determined.

Results

Rasch analysis was performed on the responses given to the 43 items included in the initial questionnaire. Four items were excluded because more than 50% of respondents answered that the situation proposed did not apply to them. Seven items that showed overlapping and disordered categories were also removed. Pairwise DIF analysis were performed in subgroups defined by underlying disease and administration route. Eleven items presented DIF and were eliminated from the questionnaire. Finally, four items were deleted after analyzing the fit statistics, three of which did not fit the Rasch model and one did not belong to either of the dimensions. The final version of NutriQoL® includes 17 items.

Conclusions

NutriQoL® is a useful instrument to assess the HRQoL of HEN patients with any disease and any administration route.

Estimating Skin Cancer Risk: Evaluating Mobile Computer-Adaptive Testing

Background

Response burden is a major detriment to questionnaire completion rates. Computer adaptive testing may offer advantages over non-adaptive testing, including reduction of numbers of items required for precise measurement.

Objective

Our aim was to compare the efficiency of non-adaptive (NAT) and computer adaptive testing (CAT) facilitated by Partial Credit Model (PCM)-derived calibration to estimate skin cancer risk.

Methods

We used a random sample from a population-based Australian cohort study of skin cancer risk (N=43,794). All 30 items of the skin cancer risk scale were calibrated with the Rasch PCM. A total of 1000 cases generated following a normal distribution (mean [SD] 0 [1]) were simulated using three Rasch models with three fixed-item (dichotomous, rating scale, and partial credit) scenarios, respectively. We calculated the comparative efficiency and precision of CAT and NAT (shortening of questionnaire length and the count difference number ratio less than 5% using independent t tests).

Results

We found that use of CAT led to smaller person standard error of the estimated measure than NAT, with substantially higher efficiency but no loss of precision, reducing response burden by 48%, 66%, and 66% for dichotomous, Rating Scale Model, and PCM models, respectively.

Conclusions

CAT-based administrations of the skin cancer risk scale could substantially reduce participant burden without compromising measurement precision. A mobile computer adaptive test was developed to help people efficiently assess their skin cancer risk.