There is little evidence-based guidance to facilitate design decisions for definitive trials that evaluate treatment efficacy for adults with tinnitus. Our recent systematic review of intervention studies for tinnitus determined the current status of trial designs by identifying and evaluating the reporting of outcome domains and instruments for publications between 1st July 2006 and 12th March 2015 [1]. Eight studies published in International Tinnitus Journal contributed to the systematic review, from a total of 228. Two were designated as randomised controlled trials [2,3], four were controlled but not fully randomised [4-7], and two had a non-randomised before-and-after design [8,9]. Patient groups and interventions were diverse, as were the tinnitus-related outcome measurements, precluding direct comparison across findings. For example, main patient-reported outcome measures across the eight studies were Tinnitus Handicap Inventory, Tinnitus Questionnaire (TQ, English and German versions), the Klockhoff assessment, Tinnitus Handicap Questionnaire and visual analogue scales of tinnitus loudness and annoyance. All studies reported the treatment-related change on these measures according to whether or not they reached statistical significance, but only Argstatter et al. [4] sought to additionally interpret whether or not the observed treatmentrelated improvements were clinically significant. They reported that neuro-music therapy is statistically significant and clinically effective, with 85% of patients having improved at the end of about 8 hours of neuro-music therapy. Reporting of clinical significance is welcomed because findings can be statistically significant, without being clinically significant. Clinical significance usually means that a treatmentrelated change corresponds to a noticeable, appreciable difference that is of value to the patient and healthcare professional, and cannot be attributed to chance. Clinical significance is thus important because it tells us whether a trial finding is meaningful to patients, and also helps healthcare professionals to interpret the research data in order to inform their evidence-based practice.

My co-authors and I learned a lot of valuable lessons about trial design during the process of systematic review process. Some of these reveal a few uncomfortable truths about how the concept of clinical significance has been handled by tinnitus investigators. In this letter, I explain what these uncomfortable truths are and I also suggest a further incremental analysis that could be implemented on the data collected by Argstatter et al. [4], so that greater knowledge about clinical significance can assist the tinnitus community in designing and reporting good quality trials. Argstatter et al. used the TQ, and this is in common use in clinical trials [1], especially in Germany where a German version is available [10]. Across the 228 trials in our systematic review [1], 17 trials identified treatment responders using criteria defined by a pre-post change score on the TQ and/or mini-TQ (an abbreviated version of the German TQ [11]).

The first uncomfortable truth is that investigators don’t agree on what change score a patient needs to show in order for their tinnitus condition to be considered ‘improved’ on the TQ and/or mini-TQ. All criteria are reported in (Table 1). In summary, investigators defined improvement as an absolute reduction of a predetermined number of points, a percentage reduction or a change that crossed a category boundary.

Table 1. Summary of numerical criteria for what constitutes a clinically meaningful change on versions of the TQ, using investigator-reported data extracted from 228 trials¹.

The second uncomfortable truth is that investigators typically don’t involve patient viewpoints when deciding what the clinically important difference score should be. For example, Argstatter et al.4 chose their criterion based on the statistical formula (1.96*- √ 2 *standard error) using data from the mini-TQ. While this calculation accounts for the measurement properties of the questionnaire, it does not tell us how the change score relates to the patient experience of tinnitus.

There isn’t any single method for determining what the minimal clinically important difference should be. But one simple approach is an anchor-based method. This assesses what change on the TQ corresponds with a minimal important change defined on the anchor (an external criterion is used to operationalise a relevant or important difference). Typically the anchor is a single question that asks patients to rate the change in the severity of their condition on a point scale. The degree of change between the response categories ‘minimally improved’ and ‘no change’ is a good indicator of the smallest change in TQ scores that is important from the patients’ perspective. Serendipitiously, Argstatter et al. [4] did ask patients about their global impression on discharge from the neuro-music therapy groups. The question was “Was your tinnitus percept changed by the music therapy?”, with response options being “worsened”, “unaffected” and “improved”. Hence, a first approximation of the anchor-based method to determine the clinically important difference on the TQ could be calculated from these data reported in the International Tinnitus Journal.

Methodologists generally recommend triangulat ing the results of multiple methods for determining clinically important difference. Other methods include distribution-based methods such as the Standard Error of Measurement (SEM) and Smallest Detectable Change (SDC). With 151 patients allocated to intervention, Argstatter et al.4 have sufficient data to make these computations and to make a recommendation to the tinnitus community about the criterion for clinical significance of the TQ change score. This could greatly assist investigators at the design stage of the trial, and enhance consistency across studies.

References

1. Hall DA, Haider H, Szczepek AJ, Lau P, Rabau S, Jones-Diette J, et al. Systematic review of outcome domains and instruments used in clinical trials of tinnitus treatments in adults. Trials. 2016;17(1):270.
2. Amanda B, Manuela M, Antonia M, Claudio M, Gregorio B. Posturography measures and efficacy of different physical treatments in somatic tinnitus. Int Tinnitus J. 2010;16(1):44-50.
3. Robinson SK, Viirre ES, Bailey KA, Kindermann S, Minassian AL, Goldin PR, et al. A Randomized Controlled Trial of Cognitive­ Behavior Therapy for Tinnitus. Int Tinnitus J. 2008;14(5):119-26.
4. Argstatter H, Grapp M, Plinkert PK, Bolay HV. Heidelberg Neuro-Music Therapy for chronic-tonal tinnitus - treatment outline and psychometric evaluation. Int Tinnitus J. 2012;17(1):31-41.
5. Korres S, Mountricha A, Balatsouras D, Maroudias N, Riga M, Xenelis I. Tinnitus Retraining Therapy (TRT): outcomes after one-year treatment. Int Tinnitus J. 2010;16(1):55-9.
6. Latifpour DH, Grenner J, Sjodahl C. The effect of a new treatment based on somatosensory stimulation in a group of patients with somatically related tinnitus. Int Tinnitus J. 2009;15(1):94-9.
7. Lugli M, Romani R, Ponzi S, Bacciu S, Parmigiani S. The windowed sound therapy: a new empirical approach for an effective personalized treatment of tinnitus. Int Tinnitus J. 2009;15(1):51-61.
8. Mahboubi H, Ziai K, Djalilian HR. Customized web-based sound therapy for tinnitus. Int Tinnitus J. 2012;17(1):26-30 .
9. Sutbas A, Yetiser S, Satar B, Akcam T, Karahatay S, Saglam K. Low-cholesterol diet and anti-lipid therapy in managing tinnitus and hearing loss in patients with noise-induced hearing loss and hyperlipidemia. Int Tinnitus J. 2007;13(2):143-9.
10. Goebel G, Hiller W. The tinnitus questionnaire. A standard instrument for grading the degree of tinnitus. Results of a multi-center study with the tinnitus questionnaire. HNO 1994;42(3):166-72.
11. Hiller W, Goebel G. Rapid assessment of tinnitus-related psychological distress using the Mini-TO. Int J Audiol. 2004;43:600-4.