Performance of the Tinnitus Functional Index as a diagnostic instrument in a UK clinical population

The Tinnitus Functional Index (TFI) has been optimised as a diagnostic tool for quantifying the functional impact of tinnitus in US veteran and civilian groups. However, the TFI has not been fully evaluated for use in other English-speaking clinical populations despite its increasingly popular uptake. Here, a prospective multi-site longitudinal validation study was conducted to evaluate psychometric properties relevant to the UK clinical population. Guided by quality criteria for the measurement properties of health-related questionnaires, we specifically evaluated three diagnostic properties relating to the degree to which the TFI (i) covers the eight dimensions proposed to be important for diagnosis, (ii) reliably distinguishes individual differences in severity of tinnitus, and (iii) reliably measures the functional impact of tinnitus. We also examine whether clinically meaningful interpretations of the scores can be produced for the UK population.

Twelve National Health Service audiology clinics across the UK recruited 255 tinnitus patients to complete questionnaires at four time-intervals, from initial clinical assessment and then over a nine-month period. Patients completed the TFI, the Tinnitus Handicap Inventory (THI), tinnitus case history questions, a Global rating of Perceived Problem with tinnitus and a Clinical Global Impression of perceived change in tinnitus. Baseline TFI data were used to examine the factor structure, construct validity and interpretability of the TFI. Follow-up TFI data were used to examine reliability.

Confirmatory factor analysis suggested that of the eight subscales (factors) initially established for the TFI, the 'Auditory' subscale did not contribute to the overall construct 'functional impact of tinnitus', and a modified seven-factor model (TFI-22) better fit the variance in the patient scores. Both the global 25-item TFI and the global TFI-22 scores showed exceptionally high internal consistency (α ≥ 0.95), high construct validity with the THI (r = 0.80) and high test-retest reliability (ICC = 0.87). Test-retest agreement however was only deemed to be borderline acceptable (89%). Receiver Operator Characteristic analysis indicated the 25-item TFI and TFI-22 has excellent ability to distinguish between different levels of impact (Area under the curve > 0.7).

The TFI was confirmed to cover multiple symptom domains, measuring a multi-domain construct of tinnitus, and satisfies a range of psychometric requirements for a good clinical measure, including having excellent reliability, stability over time and sensitivity to individual differences in tinnitus severity. However, a modified seven-factor structure without the Auditory subscale (TFI-22) is recommended for calculating a global composite score for UK patients. Using patients' experience and Receiver Operator Characteristic analysis, a grading system was presented which identifies the distinct grades of tinnitus impact in the UK clinical population that is broadly comparable to the US-based system.