Biofeedback Rehabilitation to Improve Speaking and Eating in Public

Overview

Tongue cancer requires resection and reconstruction that can leave patients disabled with respect to speaking and eating. Impairment of tongue function can have significant impact on social interaction and employment. To try and improve speaking and eating in public, the team at the University Health Network is going to use a special device that is designed to help the patient rehabilitate after tongue cancer treatment. The research part of this study is to use a special mouthpiece or mold that is like a partial plate for upper dentures that will fit on the roof of the mouth. This mold is embedded with 62 sensors that will allow patients to visualize their tongue position during speech. With the help of a speech therapist, patients can learn different tongue positions important speaking and eating. This whole process is called biofeedback. The visualization of the tongue gives the patient the flexibility to practice at home to and learn different tongue positions important to eating and speaking. In addition, the device will provide the speech pathologist with an opportunity for easier and more precise assessment of the patient's progress which can reduce the need for visits to the hospital. The investigators expect the biofeedback training to improve the patient's speaking and eating. The study hypothesis is that the biofeedback device will improve speech intelligibility, eating in public, speaking in public, and oral intake scores.

Subjects will be enrolled in either arm of the study for duration of 4 months. Patients that are randomized into study arm 1 will come to the clinic for Speech Language Pathologist (SLP) directed face-to-face rehabilitation for a period of 2 months. The patients will then transition to participate in weekly face-to-face Electropalatography (EPG) biofeedback training. The EPG biofeedback will be used for 2 months. Patients that are randomized into study arm 2 will first undergo a self-directed tongue strengthening program (oral exercises) for a period of 2 months and then transition into EPG biofeedback training. Although randomization will be used, this is a nonstandard trial because all patients will receive biofeedback EPG. Patients from both study arms will receive biofeedback training after undergoing an SLP directed face-to-face rehabilitation program versus a self-directed tongue strengthening program (oral exercises). The SLP directed face-to-face rehabilitation program or self-directed tongue strengthening program will be initiated at least 2 weeks post-surgery. If the patient is treated with adjuvant radiation, the patient will take a self-directed biofeedback holiday with a target of restarting at least 2 after the completion of adjuvant treatment. This design was chosen because there is equipoise with respect to the efficacy of the intervention and uncertainty with respect to whether the intervention is most valuable following an SLP directed face-to-face rehabilitation program versus a self-directed tongue strengthening program. The design offers all patients the opportunity to participate in the intervention. This approach will also allow us to analyze between subject change and within subject change. The A.I.D.S assessment has been in place for approximately 50 years. The mean A.I.D.S. efficiency score (95%CI) in the current pilot study (n=16) is 0.54 (46.0, 62.3). With 33 patients in each group, there will be over 80% power to detect a 10% difference in the speech efficiency score even under the condition that there may be low (rho\<0.1) correlation within subject. Higher correlation within subject on the various scores would serve to increase the efficiency of the analysis. Bivariate comparison will be made between the arms with respect to the primary and secondary outcome variables at each time point. Linear mixed model will be used to account for multiple measures at multiple time points. The first step will be to compare arms with respect to the primary and secondary outcome variables. For those tests that are significant a backward selected regression will be performed to control for confounders and to look for important effect modifier variables. Variables related to rehabilitation participation will be modeled in the backward selected regression.