Diagnostic Accuracy of The Water Immersion Wrinkle Test for Small Fiber Neuropathy

This is a prospective, single-center, observational diagnostic-accuracy study designed to estimate the performance of the Water-Immersion Wrinkle Test (WIWT) for identifying small-fiber neuropathy (SFN) in adults evaluated by a neuromuscular department. The composite reference standard is a specialist clinical assessment consistent with SFN together with abnormality on at least one validated clinical scale (Utah Early Neuropathy Scale [UENS] or modified Toronto Clinical Neuropathy Score [mTCNS]). Quantitative Sensory Testing (QST) and Sympathetic Skin Response (SSR) are obtained as comparator tests but do not form part of the reference standard. The study adheres to STARD principles for diagnostic-accuracy research and uses predefined operating procedures for test execution, data capture, and analysis.

Study setting and workflow Participants are evaluated in the neuromuscular clinic of a tertiary hospital. The study uses a baseline-only design: all index and comparator tests are completed on the same day whenever feasible. A separate reliability session for image scoring occurs 48 hours to 7 days later without additional patient visits. The overall workflow is: (1) confirm eligibility and obtain consent; (2) complete standardized neurological examination and clinical scales; (3) perform or verify QST and SSR (irrespective of results); (4) perform WIWT; (5) acquire standardized fingertip photographs; (6) independent scoring of images by trained raters; (7) data entry and validation; (8) statistical analysis according to the prespecified plan.

Reference standard (composite) The reference standard integrates the treating neuromuscular specialist's judgment (history and examination) with at least one abnormal clinical scale among UENS or mTCNS using their published scoring rules. The scales are administered per their manuals by personnel trained on test instructions and scoring conventions. The reference standard classification (SFN vs no SFN) is recorded in a dedicated adjudication form. Test operators for WIWT, QST, and SSR do not have access to the reference classification at the time of testing.

Index test: WIWT-technical conduct Participants are seated upright. Both hands are immersed, at least to the distal interphalangeal crease, in water for 15 minutes. Before immersion, the skin is verified to be clean and completely dry. Hands are kept relaxed (neutral finger posture) and fingertips must not press against container surfaces. Because a thermostatically controlled bath is not available, immersion begins at 43-44 °C with expected passive cooling of approximately 2 °C every 5 minutes, targeting an end temperature of about 37-38 °C. Water temperature is measured with the service thermometer at 0, 5, 10 and 15 minutes to confirm the intended range. No water is added or replaced during immersion. A calibrated stopwatch is used to time the procedure. At completion, hands are gently patted dry and standardized photographs are obtained of the palmar distal phalanges of digits 2-5 of both hands. Environmental conditions are kept stable (indoor room, typical ambient temperature, indirect lighting) and recorded on the source form.

Image acquisition and standardization Photographs are taken with the clinic's digital camera or smartphone following a fixed template: (a) distance approximately 20-25 cm from fingertips; (b) camera aligned perpendicular to the palmar surface to minimize parallax; (c) neutral, diffuse illumination avoiding hard shadows; (d) inclusion of a 1-cm scale marker within the frame; (e) focus lock on the distal pad of each fingertip. Raw images are immediately reviewed for motion blur or framing errors and repeated if necessary before the participant leaves. Images are stored as high-resolution JPEG or HEIC (≥8 MP), transferred to the secure institutional drive, and automatically renamed by a deterministic convention (StudyID_Hand_Side_Digit_Timepoint.ext). No patient identifiers appear in filenames or image content.

Scoring rubric and rater training Wrinkling is graded on a 0-4 ordinal scale for digits 2-5 of each hand by trained raters who are not involved in the clinical reference assessment and are mutually blinded to QST/SSR results. Raters complete a short training using a curated set of example photographs with consensus labels, followed by a calibration exercise (≥20 images) with feedback until predefined agreement thresholds are met. For each participant, per-hand sums are computed and a bilateral summed score is derived; the bilateral classification threshold uses the prespecified rule (abnormal < 24; normal ≥ 24). For the reliability substudy, four raters independently score all images twice, separated by 48 hours to 7 days, with the second session presented in a randomized order to reduce recall. The image set used for the second session is identical to the first and contains no new photographs.

Comparator tests (QST and SSR) QST includes thermal detection thresholds and vibration thresholds measured with the clinic's standard equipment following manufacturer instructions and local laboratory SOPs, with device daily checks documented in the logbook. SSR is recorded from palmar electrodes with standard filter settings and sweep parameters; amplitude and latency are extracted according to laboratory conventions. Operators of QST and SSR are independent from the WIWT team and do not access the WIWT photographs or scores.

Bias mitigation and blinding To limit review and incorporation biases, the reference standard is defined a priori and does not include WIWT, QST, or SSR results. The WIWT examiner is masked to the reference classification and to QST/SSR readings; QST and SSR operators are masked to each other and to WIWT; image raters receive only coded image sets stripped of metadata. The order of WIWT versus QST/SSR is not fixed and is recorded to explore potential order effects. All analysis code and decision rules (for exclusions, tie handling, and outlier management) are prespecified.

Data collection and management Data are captured on paper case-report forms (CRFs) and transcribed into a REDCap-like electronic database hosted on the institutional server with user authentication, role-based access, and full audit trails. Double-data entry is applied to primary variables (reference classification, WIWT summed score, UENS/mTCNS totals); discrepancies are resolved by a third reviewer using the source documents. Automated data-validation rules check ranges (for example, temperature values at 0/5/10/15 minutes, test duration = 15 ± 1 min, score ranges 0-4 per digit), internal consistency (for example, bilateral sum equals the sum of component digits), and logical constraints (for example, no missing both "start temperature" and "end temperature"). A data dictionary defines variable names, formats, allowable codes, derivations, and edit checks. Versioning is handled at the database level; all changes after database lock require change-control approval and generate an audit entry.

Sample size rationale The planned sample size is approximately 30 participants, selected to provide acceptable precision for sensitivity and specificity estimates under plausible operating characteristics (for example, sensitivity ~0.80 and specificity ~0.70) with two-sided 95% confidence intervals of roughly ±0.10 under anticipated case-mix. The final sample size also accommodates the reliability substudy, ensuring that the number of images and raters provides adequate precision for intraclass correlation coefficients (ICCs) with confidence-interval widths ≤0.15 under expected variance components.

Statistical analysis overview Analyses are prespecified. For the primary diagnostic-accuracy analysis, WIWT is evaluated against the composite reference standard. Sensitivity, specificity, and their exact or 95% confidence intervals are reported at the prespecified bilateral sum threshold. ROC analysis treats the bilateral summed WIWT score as a continuous predictor; AUC and 95% CI are computed using non-parametric methods. Positive and negative predictive values are reported at the observed prevalence. Inter-rater reliability for the bilateral summed score uses a two-way random-effects, absolute-agreement ICC [ICC(2,1)], and intra-rater reliability uses a two-way mixed-effects ICC [ICC(3,1)] with 95% confidence intervals derived from ANOVA variance components; secondary reliability metrics (weighted kappa for ordinal per-digit grades) may be reported descriptively. Prespecified sensitivity analyses include: (a) exclusion of protocol-deviant WIWT runs (temperature/time deviations); (b) alternative temperature-window definitions; and (c) alternative classification cut-points explored descriptively without redefining the primary analysis. No hypothesis testing for between-test differences is required for the primary objective; if performed, comparisons of AUCs will use DeLong's method.

Data governance, confidentiality, and archiving All study data are de-identified at the point of storage; linkage files mapping StudyID to personal identifiers reside in a separate restricted directory accessible only to the clinical team. Photographs do not contain faces or unique marks beyond fingertip skin; rings and nail polish are removed when possible. Data are retained for the period required by institutional policy and local regulations. Access to the final analytic dataset and code is restricted to the analysis team; requests for de-identified data after publication follow the data-sharing statement entered elsewhere in the record.

Risk assessment and safety reporting WIWT, QST, and SSR are non-invasive procedures with minimal risk (primarily transient discomfort or skin maceration). Adverse events related to testing are expected to be rare and mild; any unexpected events temporally related to test execution are documented in the source forms and reported per institutional policy. The study does not use a formal data monitoring committee given its minimal-risk, non-interventional design.

Study timeline Recruitment is ongoing. Baseline testing and image acquisition occur on the day of enrollment; reliability scoring occurs within 48 hours to 7 days. Database lock is planned shortly after the anticipated primary completion date. The record will be updated as major milestones are reached (for example, changes in enrollment target, completion dates, or analysis plan).