Validity and reliability of the reflux symptom index (RSI)

Laryngopharyngeal reflux (LPR) is present in up to 50% of patients with voice disorders. Currently, there is no validated instrument that documents symptom severity in LPR. We developed the reflux symptom index (RSI), a self-administered nine-item outcomes instrument for LPR. The purpose of this investigation was to evaluate the psychometric properties of the RSL For validity assessment, 25 patients with LPR were evaluated prospectively before and six months after b.i.d. treatment with proton pump inhibitors (PPI). Each patient completed the RSI as well as the 30-item voice handicap index (VHI). For reliability assessment, the study patients were given the RSI on two separate occasions before the initiation of treatment. Normative RSI data were derived from 25 age-matched and gender-matched controls taken from an existing database of asymptomatic individuals without any evidence of LPR. The mean RSI ( standard deviation) of patients with LPR improved from 21.2 ( +/- 10.7) to 12.8 ( +/- 10.0), and the mean VHI improved from 52.2 ( +/- 24.7) to 41.5 ( +/- 25.0) after 6 months of therapy (p = 0.001 and 0.065, respectively). Of the three VHI subscales (emotional, physical, functional), only the functional subscale improved significantly (p = 0.037). Patients who experienced a five point or better improvement in RSI were I I times more likely to experience a five-point improvement in VHI (95% confidence interval = 1.7, 76.8). For reliability assessment, the first and second pretreatment RSIs were 19.9 ( +/- 11.1) and 20.9 ( +/- 9.6), respectively (correlation coefficient = 0.81, p < 0.001). The single-item correlation coefficients ranged from 0.41 to 0.91 (p < 0.05 for all items). The mean pretreatment RSI in patients with LPR was significantly higher than controls (21.2 versus 11.6; p < 0.001). The mean RSI of patients with LPR after 6 months of PPI therapy approached that of asymptomatic controls (p > 0.05). The RSI is easily administered, highly reproducible, and exhibits excellent construct and criterion-based validity.