I J Deary
Professor of Differential Psychology
University of Edinburgh, Scotland

The study I should like us to consider was carried out in what might be termed - North Britain, namely the South of Scotland, and the North of England. The study began 5 years ago, in the Royal Infirmary in Glasgow, Scotland, where I then worked. The project was supported by Scottish Office Home and Health Department.

The laryngologist principally involved in recruitment was my close colleague Ken MacKenzie. The therapist responsible for treating patients was Cam Sellars. Data analysis was aided by Dr Paul Carding, who is the Senior Lecturer in Voice Pathology. Paul and I now work together in the Otolaryngology department in the university of Newcastle upon Tyne. We share research projects and also a typical UK Voice clinic, where we jointly assess certain categories of dysphonic patient.

The study was truly interdisciplinary, also having input from Prof Ian Deary, who works also on the East Coast of Britain, but North of the Scottish border, in Edinburgh, once called the Athens of the North. Ian has helped us to study a number of ORL psychosomatic conditions such as globus sensation and temporomandibular joint dysfunction. We feel that this quality of psychological input is vital in the assessment of dysphonic patients in the UK.

Dysphonia (hoarseness) represents the interaction of pathophysiological, emotional and psychological factors. UK otolaryngologists see at least 48,000 new dysphonic patients per annum. The commonest cause of dysphonia is functional dysphonia. Over half of our new referrals will undergo speech and language therapy- at least 25,000 annually, yet there is little Level 1 evidence for its efficacy. In other words there are very few randomised, prospective, controlled trials.

Many voice clinic patients have entered a vicious cycle: psychological factors exacerbate voice pathology, while poor voice quality adversely affects psychological well-being. Conventional speech and language therapy aims to

• improve voice quality
• enhance psychological well-being
• relieve the functional and social disabilities.

Our study objective was to provide the first major randomised comparison of voice therapy with the natural history of dysphonia.

• to assess the efficacy of voice therapy for dysphonia
• to establish which patient subgroups may benefit most
• to compare the impact of treatment on different types of outcome measure, including quality of life

We recruited consecutive outpatients attending the Department of Otolaryngology, Head and Neck Surgery, Glasgow Royal Infirmary, with a primary complaint of dysphonia (hoarseness). present for a minimum of two months and without any significant pathology (e.g. polyp, papilloma, tumour, vocal cord palsy) or need for surgery.

Inclusion criteria: Age > 16 years old, motivation to resolve the voice problem, willingness to enter into regular voice therapy sessions.

Exclusion criteria: Previously treated dysphonia, neurological disease or upper aerodigestive tract malignancy; marked hearing impairment; acid reflux; multiple medical complaints; performing voice user requiring urgent intervention; puberphonia; and transsexual

We recruited 204 patients (51 men, 153 women).

By study completion 12 - 14 weeks later, approximately 33% of subjects had dropped out or been excluded, leaving 63 patients in the treatment group and 70 patients in the observation group

Four groups of outcome measures were used

Patient self - report The patient’s view of the impairment has a major influence on response to treatment.

Patients rated their own voice quality using the validated Vocal Performance Questionnaire, developed by Paul Carding. This has 5 point scoring on 12 items (1-5, 5 = worst), the maximum ( most dysphonic) VPQ score thus = 60.

Observer rating A digital analogue tape recording of the patient’s reading of the "Rainbow" passage which is a standard phonetically balanced test, was analysed by a speech therapist, blind to treatment group and status. Ten aspects of voice including laryngeal tone, pitch and resonance were rated 0-5 on the Buffalo Voice Profile scale, whose "overall rating" item was chosen as the key observer-rated variable.

Computer derived (objective) acoustic variables were also extracted by the blinded therapist. Two key variables are presented - "jitter", pitch perturbation, and "shimmer", amplitude perturbation - using the Computerised Speech Laboratory (Model 4300B). Kay Elemetrics Corp: New Jersey, USA. The higher the score on these two variables, the more dysphonic the voice.

II - Laryngoscopic features were assessed by an otolaryngologist at videorecorded flexible nasolaryngoscopy (Olympus ENF 3) using on a 4-point (0-3) rating scale of 4 features - nodule formation, laryngitis, glottic escape and hyperfunction of the laryngeal musculature.

III - Psychological Measures Numerous psychological variables are well - known to be abnormal in dysphonics. These findings may be nonspecific attributes of functional disorders in general, yet are also important because they may explain differences between patients and observers. Two key measures of psychological distress are reported.

The Hospital Anxiety and Depression Scale (HADS) is a self-completed questionnaire which assesses recent anxiety and depression, and provides an overall score for each. The Clinical Interview Schedule (CIS-R) was conducted by a trained psychologist. The interviewer rates 14 aspects of non-psychotic psychiatric disturbance including sleep problems, mood disturbance and somatic symptoms.

IV - Quality of Life was assessed by the Short-Form-36 (SF-36), an internationally validated general health status measure with which many of you will be familiar.

The SF-36 is self-administered and takes less than 10 minutes to complete. Its 8 subscales quantify a broad range of health issues. High scores on the SF-36 indicate better health.

At entry to the study baseline data were recorded and the subjects then all seen by Mr Sellars, the treating therapist. He obtained a number for random allocation of subjects to either a course of voice therapy or a period of observation. Random numbers were supplied by an independent worker in a separate department. The voice therapy was also all delivered by CS according to a protocol derived from our review of the type, form and frequency of UK speech therapy. Mr Sellars was not involved in the collection, storage or analysis of data in any way. At no time did he convey therapy status to any other member of the research team. The therapy group initially comprised 100 patients. Eight subjects were excluded for a range of reasons e.g. cognitive inability to cope with the protocol. A further 8 subjects failed to attend the first therapy appointment. A further 10 subjects failed to attend during the course of therapy. Thus 74 patients completed a course of voice therapy.

After 6 weeks of therapy or observation, all measures except the Clinical Interview Schedule were recorded again in full on the 145 patients (71%) who continued to this stage.

Following another 6-8 weeks, all measurements were repeated again on the 133 (65%) who completed the study. Comparisons were made at the 3 time points of the 70 who completed treatment with the 63 in the no treatment group.

Speech Therapy Intervention:

A routine voice history was taken and a further recording of the "Rainbow" passage made for possible reference in therapy. In line with current practice, 11 out of the 73 who completed treatment received advice regarding good vocal hygiene and optimal voice production and did not proceed to a fuller programme of voice therapy. (This option did however remain open to them until attendance for Visit 2 of the study.)

The remaining 62 subjects underwent individualized programmes of voice therapy for 2-8 treatment sessions. Treatment reflected current effective practice and was characterised a) in units of 5 minutes b) according to 20 treatment types. In the final session, a review recording of the "Rainbow" passage was made.

A mixed model Analysis of Covariance (ANCOVA) was used. Therapy was used as a between subjects factor (voice therapy versus no treatment). Time (only visits 2 and 3) was a repeated measure. The baseline values of the relevant outcome variables were used as covariates. This allowed a direct test of the main effect of therapy. Only data which were complete for all three time points were analysed for each variable.

To establish separately the efficacy of voice therapy at treatment completion and follow-up, post hoc tests were conducted on all variables which showed significant effects of therapy, time, or therapy x time interactions. Standardised residuals were calculated and therapy and observation groups were compared by independent sample t tests conducted on the residuals.

Analysis of laryngoscopic features: Ratings for each laryngoscopic feature for each patient at visit 2 were subtracted from those at visit 1. Ratings for each feature at visit 3 were also subtracted from those at visit 1. Subjects were then assigned a category (0,1,2) according to whether they had improved, deteriorated, or stayed the same. A series of Chi Square intergroup comparisons was performed.

Quality of Life (SF-36) Analysis Subjects’ raw scores on each SF-36 subscale were normalized to a scale ranging from 0 (worst) to 100 (best). Dysphonic patients’ scores were compared with those of healthy controls. (Data taken from Oxford Healthy Life Survey 1991/2, HSRU, Oxford 2). Comparisons were made using unpaired t-tests with two-tailed p values.

The therapy schedules for the treated subjects were collated.

(A further 2 subjects had >6 sessions of therapy.)

The 73 treatment group subjects were 56F, 17M, mean age 51 years (SD 14). The 72 observation subjects were 50F, 22M, mean age 52 years (SD 13).

Overall the attrition rate was as expected for such a large study sample, and since there was no significant difference between the therapy and no treatment groups, it does not affect the results. Analyses were only conducted on complete data sets in each instance.