Research PaperPerformance of the Tinnitus Functional Index as a diagnostic instrument in a UK clinical population
Introduction
The experience of tinnitus involves much more than the ‘phantom’ sensation of sound since the condition can also impact on daily functioning and cause emotional distress (Henry et al., 2016, Mohamad et al., 2016, Pierzycki et al., 2016, Szczepek et al., 2014). Thus, for those who do find tinnitus bothersome, it can be described as a multi-dimensional condition. As such, it is best captured using a multi-domain patient-reported questionnaire whereby multiple items ask about particular aspects/domains of the condition which are deemed to be important (Hall et al., 2016, Henry et al., 2016). Many tinnitus questionnaires, such as the Tinnitus Questionnaire (TQ; Hallam, 2008, Hallam, 1996, Hiller and Goebel, 1992), Tinnitus Handicap Inventory (THI; Newman et al., 1996), Tinnitus Reaction Questionnaire (TRQ; Wilson et al., 1991), and Tinnitus Handicap Questionnaire (THQ; Kuk et al., 1990), have known measurement properties that are consistent with their use in clinical diagnosis i.e. good discriminative power (Kamalski et al., 2010, Kirshner and Guyatt, 1985). However, in a systematic review of the psychometric properties of tinnitus questionnaires, Kamalski et al. (2010) did not identify or report any evidence on whether authors had provided clinically meaningful interpretations of the scores. More recently, Fackrell et al. (2014) reviewed the validity, reliability, responsiveness, and interpretability of tinnitus questionnaires using an internationally recognised set of criterion (Terwee et al., 2007) and reported that the evidence for the discriminative capabilities of these tinnitus questionnaires varied widely. The evidence was limited and hard to determine for content validity of the TQ, TRQ, and THI, for structural validity of the TQ, and TRQ, and for the clinical interpretation of the scores of the TQ, TRQ, and THQ (Fackrell et al., 2014). The authors concluded that, although the THQ has provided normative data, the ability to provide clinical interpretations of the scores has only been determined for the THI, with a defined established UK-based grading system. It was noted, however, that this grading system was solely based on expert opinion and the statistical properties of the scores. As such, these grades do not necessarily reflect the actual patient experience.
Importantly, the evaluation by Fackrell et al. (2014) included the Tinnitus Functional Index (TFI; (Meikle et al., 2012). First published in 2012, the TFI differs from previous tinnitus questionnaires in a number of important and positive ways; namely its careful development, comprehensive coverage of many important tinnitus complaints, interpretability of scores and responsiveness to treatment-related change (Fackrell et al., 2014). Not surprisingly, the tinnitus community at large appears eager to embrace its use. In the period 2012–2015, the TFI has established itself as the second most commonly used tinnitus questionnaire in UK National Health Service (NHS) tinnitus services; the THI is most commonly used (Hoare et al., 2015). However, it is important for our communities to appreciate that the statistical properties of the TFI are not immutable. Whilst it might be valid, reliable, and interpretable in one target population, it may behave in quite a different way in a different population (e.g. Streiner et al., 2014). As the TFI gains in international popularity in the clinic, it is important that its discriminative properties be evaluated thoroughly for each new setting and population.
It is well documented that the TFI was developed using data collected in the US, some in specialist tinnitus clinics but principally in Veteran's Affairs (VA) hospitals (58% of patients) (Meikle et al., 2012). In VA hospitals, those patients tend to be male, with an active military background, potentially experiencing a range of service-related co-morbidities, and their tinnitus is considered as a service-related condition which may entitle them to compensation. This rather unique provenance of the TFI warrants caution in terms of how well those psychometric properties transfer to different target populations.
Since the development of the TFI (Meikle et al., 2012), several evaluations of the questionnaire have been conducted in English speaking and non-English speaking countries. These evaluations increase our understanding and optimising the use of this questionnaire for research and clinical practice alike. To date, the American-English version of the TFI has been evaluated in US Veterans (Henry et al., 2016), a general clinical population in New Zealand (Chandra et al., 2014) and a research population drawn from the general public in the UK (Fackrell et al., 2016). The psychometric exploration reported by Henry et al. (2016) has the same potential limitation (not generalizable) as was noted in the original development study (Meikle et al., 2012). Fackrell et al. (2016) raised some doubts of the stability of the 8-factor structure of the TFI when used in a UK-based research population, namely that the auditory subscale appeared not to contribute to the measure of global functional impact of tinnitus. There have been four independent evaluations in different target populations, where the TFI has been translated into Dutch (Rabau et al., 2014), Swedish (Hoff and Kähäri, 2016, Müller et al., 2016), and Polish (Wrzosek et al., 2016). In general, evaluations of these translated versions showed the TFI to have good discriminative properties. However, there was also some uncertainty over its proposed factor structure. In all of those studies, Exploratory Factor Analysis (EFA) was conducted which identified different patterns in the data, typically with only five or six factors initially identified, although all reported forced eight-factor models as being satisfactory (Rabau et al., 2014, Hoff and Kähäri, 2016, Müller et al., 2016, Chandra et al., 2014). Only the Polish study included Confirmatory Factor Analysis (CFA) to test the proposed eight-factor structure, finding it to be unsatisfactory (Wrzosek et al., 2016). Instead, their EFA indicated that a five-factor solution best explained the Polish population data. Interpretability was not assessed in any of those studies.
Meikle et al. (2012) have proposed interim grading systems for the TFI, but the question of whether this interpretability of the global scores, an essential requirement for the suitability of a questionnaire in clinical practice or research, is transferable to other populations is yet to be addressed in any subsequent psychometric evaluation.
In the present study, we examined the psychometric properties of the TFI for a large clinical sample of UK NHS patients treated for tinnitus. In designing this study we were guided by quality criteria for the measurement properties of health-related questionnaires as outlined by Mokkink et al. (2012) and Terwee et al. (2007). Unlike our previous work (Fackrell et al., 2016), this study was specifically designed to evaluate the TFI as a reliable and valid measure of tinnitus severity for use in a tinnitus clinical population, and to determine its responsiveness and interpretability. This study is particularly important because it is based on a study sample drawn from a general (i.e. non-military) help-seeking clinical population.
The aims of the study were to evaluate the degree to which the TFI:
- i)
covered the proposed eight important dimensions of tinnitus-related impact,
- ii)
reliably distinguished one patient from another,
- iii)
reliably measured the impact of tinnitus,
- iv)
produced a grading scheme that can give a meaningful diagnostic interpretation to the UK clinical population
Section snippets
Materials and methods
This was a prospective multi-site, repeated-measures validation study. Ethical approval was granted by Cornwall and Plymouth Research Ethics Committee (13/SW/0234), and Nottingham University Hospitals NHS Trust was Sponsor.
Analysis methods
The methodological approach taken here was underpinned by Classical Test theory principles, in which a person's “true score” is directly unobservable. Every observed score is assumed to be made up of measurement error and the person actual “true” attitude or attribute on the latent construct that is being measured, in this case tinnitus (Raykov and Marcoulides, 2011). The criteria for acceptable psychometric properties described below were guided by established frameworks to evaluate
Patient characteristics
A total of 255 tinnitus patients (male: 149 (59%), female: 105 (41%)) were enrolled and completed T0 measurements. The average age was 53.6 years (SD = 13.4) with a range of 18–84 years. Just under 50% of patients had experienced tinnitus for less than 2 years, 30% reported tinnitus duration between 3 and 10 years, and the remainder reported experiencing tinnitus for more than 11 years. Descriptors of tinnitus sounds included whistling, buzzing, ringing, hissing, clicking, cracking, whooshing,
Discussion
The current study provides the first independent and comprehensive psychometric evaluation testing the diagnostic utility of the TFI in a UK clinical population, building on our previous psychometric evaluation in a UK tinnitus research volunteer population (Fackrell et al., 2016). Notably, we conclude for a UK clinical population that although the TFI proposed by Meikle et al. (2012) generally produced a reliable diagnostic tool with good discriminative properties, and good convergent validity
Acknowledgements
The authors would like to thank the following NHS clinical facilities and individual clinicians, who worked tirelessly to recruit all the patients to the study: Tony Kay, Audiology services, Aintree University Hospital NHS Foundation Trust (Liverpool), Mary Mitchell, Hearing, Tinnitus & Balance Rehabilitation Service, Belfast Health and Social Care Trust (Belfast), Saskia Harden, Audiology, Brighton & Sussex University Hospitals NHS Trust (Brighton), Rachel Knappett and David Baguley,
References (74)
On tests and indices for evaluating structural models
Pers. Individ. Dif.
(2007)- et al.
Defining clinically meaningful change in health-related quality of life
J. Clin. Epidemiol.
(2003) Receiver operating characteristic analysis: a primer
Acad. Radiol.
(2005)- et al.
Psychometric properties of the Tinnitus Functional Index (TFI): assessment in a UK research volunteer population
Hear. Res.
(2016) - et al.
Measuring change over time: assessing the usefulness of evaluative instruments
J. Chronic Dis.
(1987) - et al.
Measuring health status: what are the necessary measurement properties?
J. Clin. Epidemiol.
(1992) - et al.
Tinnitus Functional Index: development, validation, outcomes research, and clinical application
Hear. Res.
(2016) - et al.
A psychometric study of complaints in chronic tinnitus
J. Psychosom. Res.
(1992) - et al.
Measuring disease-specific health-related quality of life to evaluate treatment outcomes in tinnitus patients: a systematic review
Otolaryngol. - Head. Neck Surg.
(2010) - et al.
A methodological framework for assessing health indices
J. Chronic Dis.
(1985)