Introduction
The diagnosis of laryngopharyngeal reflux (LPR) can be made on the basis of symptoms and laryngeal findings.Reference Koufman, Aviv, Casiano and Shaw1 Ambulatory, 24-hour, double-probe (i.e. simultaneous oesophageal and pharyngeal) pH monitoring remains the ‘gold standard’ when a diagnosis of LPR is in question. As a screening tool, 24-hour, double probe pH monitoring has been criticised for its lack of sensitivity as well as its high cost and inconvenience to the patient.Reference Bilgen, Ögüt, Kesimli-Dinç, Kirazili and Bor2 A small amount of pharyngeal reflux occurs in the normal population; therefore, the interpretation of results remains controversial.Reference Merati, Ulualp, Lim and Toohill3, Reference Ford4
An initial therapeutic trial of high-dose proton pump inhibitors (PPIs) has been advocated as the first step in diagnosis and treatment of patients with suspected LPR; however, no universally accepted recommendation exists for the length of the therapeutic trial.Reference Ford4, Reference Johnson5 A minimum of four weeks is required before a diagnosis of ‘no response’ is made.Reference Johnson5 Despite the limited benefit from PPI compared with placebo in placebo-controlled trials, patients diagnosed with LPR continue to be given empirical PPI treatment.Reference Vaezi6, Reference Karkos and Wilson7 Economic analysis has revealed empirical PPI treatment to be a cost-effective diagnostic tool.Reference Williams, Szczesniak, Maclean, Cole and Cook8 In uncontrolled trials, the response of LPR patients to diagnostic one-month PPI treatment is unpredictable, ranging from 27 to 83 per cent; for three months, the response is between 41 and 100 per cent.Reference Vaezi6, Reference Karkos and Wilson7, Reference Ulualp, Toohill and Shaker9 Placebo-controlled trials have also shown conflicting results.Reference Karkos and Wilson7
Few trials have analysed the predictors of response to PPI treatment. Park et al. Reference Park, Hicks, Khandwala, Richter, Abelson and Milstein10 found that pre-therapy interarytenoid mucosa and true vocal fold abnormalities were associated with a two-fold increase in symptom response to PPI treatment. Williams et al. Reference Williams, Szczesniak, Maclean, Cole and Cook8 concluded that pharyngeal pH-meter measurements may prove useful; while other investigated factors cannot predict the response to therapy with certainly.Reference Garrigues, Gisbert, Bastida, Ortiz, Bau and Nos11
The aims of our study were to assess the benefit from one and three months of empirical PPI treatment in patients with suspected LPR, and to define pre-therapy factors which may predict success with short-term PPI treatment.
Materials and methods
One hundred and fifty-six consecutive adult patients consulting otorhinolaryngologists at the Kaunas Medical University Department of Otorhinolaryngology out-patient clinic were suspected of LPR, and considered for study enrolment, due to: chronic hoarseness (exceeding one month), throat-clearing, globus sensation, sore throat, or a cough with the typical laryngeal findings of reflux laryngitis (i.e. erythema, oedema, posterior glottis hypertrophy, vocal fold oedema and subglottic oedema).
In order to be eligible for inclusion, patients were required to have: (1) at least two laryngeal symptoms rating no less than two points on a zero to three Likert severity scale, and (2) at least posterior laryngitis, after exclusion of other possible causes of laryngeal inflammation (i.e. allergy, infection, sinus pathology and asthma). Subjects were excluded if they met the following criteria: additional laryngeal findings such as nodule, polyp, cyst or neoplasm; prior anti-reflux surgery; upper respiratory tract infections in the last month; pregnancy during the study period; and diagnosed psychiatric illness.
Eligible subjects then completed questionnaires regarding: patient demographics; medical history; allergy history; tobacco use; voice training (practical voice training for not less than two hours/week for two years); symptoms and their severity; underwent videotaped laryngoscopy or upper gastrointestinal endoscopy; voice assessment with voice range profile; and quality of life assessment measures (i.e. voice handicap index, hospital anxiety and depression scale, and general well-being assessment).
Following investigations completion, all selected patients were treated with omeprazole for three months. In the first month, patients were randomised (i.e. elementary randomisation by draw of an envelope containing the dose) to receive omeprazole 20 mg once daily or 20 mg twice daily, with a 1:3 randomisation ratio as determined by the investigators' meeting. During the next two months, all patients received omeprazole 20 mg twice daily. All patients tolerated their treatment well, and no additional medication or special orders to change habits were given.
Control assessments were performed after one and three months of therapy, comprising self-rating of symptom severity, videolaryngoscopy, voice assessment and quality of life assessment. Treatment efficacy parameters included: self-rated total symptom score, videolaryngoscopic score and general well-being assessment. The overall benefit from therapy was assessed according to the chosen criteria. Patients were considered to be responders if they had a reduction of at least 50 per cent in their total symptom score and of at least two points in their videolaryngoscopic score, and were satisfied with the results. Recovery was noted if the total symptom score improved at least 80 per cent and videolaryngoscopic findings became normal. Improvement was noted if the total symptom score improved by less than 50 per cent. No change was noted if the total symptom score improved by less than 10 per cent and patients' self-rating indicated no improvement.
The following independent variables were assessed as possible predictors for omeprazole treatment outcomes at one and three months of treatment: gender; age; active smoking; dose of omeprazole (20 mg once daily vs 20 mg twice daily for first month of treatment); baseline reflux symptom severity; videolaryngoscopic scores and endoscopic findings; overall vocal dysfunction degree (based on voice range profile); and quality of life parameters (voice handicap index, hospital anxiety and depression scale, and general well-being assessment).
Written, informed consent was obtained from all patients. The study protocol was approved by the ethics committee of the Kaunas University of Medicine, Lithuania.
Patient evaluation
Reflux symptom scores
These were self-assessed by patients, and included six laryngeal symptoms (i.e. hoarseness, throat-clearing, globus sensation, sore throat, chronic cough and choking episodes during the night) and three oesophageal symptoms (i.e. heartburn, regurgitation and odynophagia). The severity of each symptom was scored according to a four-point Likert scale, in which zero indicated no symptoms and three indicated severe symptoms (i.e. symptoms hard to tolerate and interfering with planned activities). A symptom was defined as being present if it was experienced for two or more days in a two-week period and scored more than zero points on the Likert scale. The sum of each individual symptom severity score, multiplied by the number of presenting laryngeal and oesophageal symptoms, made up the laryngeal (range, zero to 108 points) and oesophageal (range, zero to 27 points) scores; the sum of both these scores made up the total symptom score (range, zero to 135 points). This composite score was intended to enhance the power of the data, and has been used previously by the current authors and others.Reference Siupsinskiene, Adamonis and Toohill12, Reference Noordzij, Khidr, Evans, Desper, Mittal and Reibel13.)
Telescopic videolaryngoscopy
This investigation, using a Kay Elemetrics RLS (Kay Elemetrics Co., Lincoln Park, NJ, USA) equipment with 70° rigid endoscope, was performed according to a standard protocol, by the same examiner, using the same equipment, for each patient. Laryngeal abnormalities in the four laryngeal regions (i.e. posterior, vocal folds, vestibular folds and subglottic area) were evaluated by rating each region on colour, oedema and hypertrophy, using a four-point Likert scale (zero = no signs; three = severe signs), using a preprinted evaluation form. The sum of the separate region scores made up the videolaryngoscopic score (range, zero to 36 points). A score of more than two points for the posterior region was considered as positive for posterior laryngitis.Reference Siupsinskiene, Adamonis and Toohill12 Baseline and control ratings of video recordings were performed by observers blinded to the patient's identity. The intra-rater correlation for video recordings (calculated using 10 randomly selected patient video recordings over a five-month time interval) was 0.81.
Upper endoscopy findings
Endoscopy findings for the upper gastrointestinal tract were determined for each patient. The extent of oesophageal mucosal damage was assessed using the Los Angeles classification grading system. No less than A grade oesophagitis was required to confirm a diagnosis of gastroesophageal reflux disease.Reference Dent, Brun, Fendrick, Fennerty, Janssens and Kahrilas14
Quantitative voice assessment
Voice was analysed by voice range profile. This was registered in an ordinary 5 × 3 m room (with noise level not exceeding 40 dBA) in the standard manner (i.e. by ear), according to the recommendations of the Union of European Phoniatricians.Reference Pedersen, Frokjaer-Jensen, Pabst, Shutte, Hacki, Hansen and Kotby15 The pitch range was measured with the help of an electronic keyboard (Fujiyama 3A; Fujiyama Tech. Co., Shenzhen, China), within a range of four octaves. Sound pressure level was determined from a sound pressure level meter (VEB Robotron-Messelektronic, Otto Schön, Dresden, Germany), using the slow meter damping and A-weighted frequency curve dBA. Only precise corresponding sounds sustained for at least two seconds were registered. Four voice range profile parameters were analysed: pitch range, measured in semitones; maximum–minimum intensity range, measured in dBA; total voice range profile area, calculated by the square method; and high frequency area (part of the total area), calculated from 349.2 Hz for men and 523.3 Hz for women, measured in cm2. The overall vocal dysfunction degree was calculated, from zero (normal) to three (severe dysfunction), according to the original guidelines, from four assessed voice range profile parameters.Reference Šiupšinskienė16
Quality of life parameters
Voice handicap index was self-rated by the patients using a 30-item questionnaire.Reference Jacobson, Johnson, Grywalski, Silbergleit, Jacobson and Benninger17 Each statement was rated by frequency (zero = never; four = always). The voice handicap index generated a total score ranging from one to 120 and three subscale scores (functional, physical and emotional). The hospital anxiety and depression scale was self-assessed using a 14-item questionnaire with four response categories, measuring the levels of anxiety and depression in two separate subscales.Reference Zigmond and Snaith18 Scale scores ranged from zero (no symptoms) to 21 (maximum distress). A score of 11 in either of the subscales indicated probable psychological distress. General well-being was self-rated using a 100 mm visual analogue scale (VAS), from zero (extremely bad) to 100 (excellent).
Statistical methods
Statistical analysis was performed using the Statistical Package for the Social Sciences version 10 for Windows software (SPSS Inc, Chicago, Illinois, USA). Univariate analysis between responders and non-responders was performed using the t-test for normally distributed quantitative parameters, and the Mann–Whitney U and chi-square or Fisher's exact tests for non-parametric data. A level of significance of 0.05 was used. An analysis of variance model in repeated measures at three time points was used with Bonferroni correction for multiple comparisons, yielding significant p value if less than 0.001.
Multivariate analysis for prediction of response to one and three months' omeprazole treatment was performed using logistic regression analysis, initially by entering all independent variables at a single step. To stay in the model, variables were required to be significant, using a 5 per cent significance level.
These variables were then separately analysed with binary logistic regression in order to obtain their predictive effect for positive treatment response. Odds ratios (OR) are expressed with their 95 per cent confidence intervals (CIs). The Spearman correlation coefficient was used for assessing intra-rater reliability (r > 0.30 were considered to represent significant association).
Results
Study population
Of the 156 consecutive patients considered for study, 56 were excluded as a result of the exclusion criteria and/or refusal of medical treatment, leaving 100 eligible patients. The study population comprised 75 female and 25 male subjects, with a mean age of 39.9 years ± a standard deviation of 13.8 years. Eleven patients were smokers and 23 had trained voices.
All patients had two or more laryngeal symptoms and posterior laryngitis, 85 patients had vocal fold oedema, and one had granuloma. The prevalence of laryngeal symptoms was as follows: throat-clearing, 92 per cent of cases; hoarseness, 85 per cent; globus sensation, 74 per cent; sore throat, 66 per cent; chronic cough, 62 per cent; and choking, 32 per cent.
Oesophageal symptoms were present in 76 of the cases with predominant heartburn symptom (i.e. 58 per cent of patients). Oesophagitis was confirmed in 21 of the selected patients: 17 (80.9 per cent) of these patients had oesophagitis A and four (19.1 per cent) had oesophagitis B.
During the first month of treatment, 28 patients received omeprazole 20 mg once daily and 72 patients received omeprazole 20 mg twice daily. There was no significant difference in these patients' demographic characteristics, tobacco use, voice training, total symptom score or videolaryngoscopic score.
Response to therapy
After one month of treatment, analysis showed a significant improvement in the means of all tested efficacy parameters (p < 0.001) (Table I). According to the selected criteria, 56 of 100 (56 per cent) LPR patients were classified as responders: 12 of the 28 (42.9 per cent) patients receiving omeprazole once daily, and 44 of the 72 (61.1 per cent) patients receiving omeprazole twice daily (p > 0.05). Of the 56 responders, 10 (17.8 per cent) patients showed an improvement of equal to or greater than 80 per cent. An improvement of less than 50 per cent was noted in 30 (30 per cent) of the 100 LPR patients, while 14 (14 per cent) of these patients were classified as experiencing no change.
Table I Effect of omeprazole treatment at one and three months in suspected LPR patients*
Data are expressed as mean and standard deviation. The reduction in total symptom score and laryngeal scores and increase in general well-being score shows improvement.
* n = 100. †Statistically significant difference between baseline and each follow-up measurement; p < 0.001, Bonferroni correction. LPR = laryngopharyngeal reflux; mth = months
At three months of treatment, a further reduction of symptom and videolaryngoscopic scores and improvement in general well-being were achieved (p < 0.001) (Table I). Thirty-six more patients were classified as responders, taking the total to 92 of 100 (92 per cent). Of these patients, recovery was noted in 67 (72.9 per cent). Increasing the omeprazole dose to twice daily for an additional two months resulted in an improvement in total symptom score and videolaryngoscopic score in 12 of 16 (75 per cent) non-responders who received 20 mg once daily during the first month. Three-month non-responders had signs of vocal abuse, psychological distress and active smoking, and one patient had erosive oesophagitis B with hiatal hernia.
Factors that may influence treatment outcomes
Prediction of response to omeprazole after one month's treatment
After one month of treatment, 56 per cent of patients were considered to be responders because their total symptom score had improved at least 50 per cent, their videolaryngoscopic score had improved by at least two points and they were satisfied with the results. When comparing the general results of symptomatic responders vs non-responders for all treatment regimen patients combined, 12 independent pre-therapy factors were compared to determine potential predictors of therapy response (i.e. gender; age; active smoking; dose of medicine – 20 mg vs. 20 mg BID reflux symptom severity score; endoscopic findings; overall vocal dysfunction degree; voice handicap index; anxiety score by HAD scale; depession score by HAD scale; and general well-being assessment). Univariate analysis showed a significant difference between responders and non-responders regarding pre-therapy heartburn severity (p = 0.019, Mann–Whitney U test) and mean anxiety score (p = 0.036, t-test) (Table II). Patients who had entered the study with higher heartburn scores and lower anxiety scores responded significantly better to treatment. Logistic regression analysis revealed anxiety, heartburn and medication dose as relevant independent factors for the prediction of one-month treatment response. These variables are listed in Table III. An increase of one anxiety score point decreased the odds ratio for a positive response by 1.15 times (p = 0.014), while a higher heartburn score on entry increased the odds by approximately two times (odds ratio 1.99, P = 0.004) and dose of omeprazole 20 mg twice daily – by three times (odds ratio 3.05, P = 0.027).
Table II Outcome of one-month omeprazole treatment: data for possible influencing Pre-therapy factors
Quantitative parameters are expressed as mean and standard deviation (SD); non-parametric data are expressed as median and interquartile range (IQR; 25th and 75th percentiles). *n = 56; †n = 44. ‡Statistically significant difference; p < 0.05. Yr = years; VHI = voice handicap index; HAD = hospital anxiety and depression scale
Table III Factors significantly predictive of response to one-month omeprazole treatment: logistic regression analysis
Significance level, p < 0.05. B = logistic regression coefficient; OR = odds ratio; CI = confidence interval
Prediction of response to omeprazole after three months' treatment
Comparison of 12 independent pre-therapy factors for responders (n = 92) and non-responders (n = 8), after three months' omeprazole treatment, showed no statistically significant difference. With logistic regression, similarly, no statistically significant association with response was found for any of these pre-therapy factors. Of 44 non-responders after one month of omeprazole treatment, 38 (86.3 per cent) were considered to be responders at three months. A higher baseline videolaryngoscopic score had a significant influence on these patients' positive responses (p = 0.007) (Table IV), although logistic regression analysis showed no significant association for any of the pre-therapy factors analysed.
Table IV Outcome of three-month omeprazole treatment for 44 one-month non-responders; data for possible influencing Pre-therapy factors for three-month response
Quantitative parameters are expressed as mean and standard deviation (SD); non-parametric data were expressed as median and interquartile range (IQR; 25th and 75th percentiles). *n = 38; †n = 6. ‡Statistically significant; p < 0.05. Yr = years; VHI = voice handicap index; HAD = hospital anxiety and depression scale
Discussion
The association of gastroesophageal reflux with extra-oesophageal disorders is often made by symptoms and findings alone; however, ambulatory, 24 hour, two- or three-site pharyngoesophageal pH monitoring and trials of acid suppression solidify the diagnosis of Laryngopharyngeal reflux (LPR).Reference Koufman, Aviv, Casiano and Shaw1, Reference Ford4, Reference Johnson5 Treatment with Proton Pump Inhibitors (PPI) has been shown to be sensitive in up to 85.7 per cent of cases and establishes a causal relationship.Reference Bilgen, Ögüt, Kesimli-Dinç, Kirazili and Bor2, Reference Siupsinskiene and Adamonis19 The length of diagnostic PPI trials has varied from two to 12 weeks.Reference Bilgen, Ögüt, Kesimli-Dinç, Kirazili and Bor2, Reference Park, Hicks, Khandwala, Richter, Abelson and Milstein10, Reference Siupsinskiene and Adamonis19 Treatment frequently requires six months or more for optimal relief and disappearances of physical findings.Reference Koufman, Aviv, Casiano and Shaw1, Reference Ford4 Previous trials have suggested that PPI therapy should be extended for at least one month before a diagnosis of ‘non-responder’ can be made.Reference Johnson5 Identification of the characteristics of patients who would respond to PPI treatment has been indicated by recent, reliable studies.Reference Gatta, Vaira, Sorrenti, Zucchini, Sama and Vakil20 An earlier positive response to treatment has benefits for both the patient and physician. Patients who responded during the one-month period could be reassured regarding the accuracy of diagnosis and the effectiveness of administered treatment.
The aims of our study were to assess the benefit from one and three months of empirical PPI treatment in patients with highly suspected LPR, and to define pre-therapy factors which may predict success with short-term PPI treatment. Our analysis demonstrated that after one month of empirical omeprazole treatment, 56 per cent of selected patients were classified as responders, and that 92 per cent of patients responded to three months of treatment. These data are in concordance with the findings of other studies, that is: for four weeks' empirical high-dose PPI treatment, response rates of 27 to 83 per cent; and for three months' treatment, response rates of 41 to 100 per cent in uncontrolled studies and of 40 to 90 per cent in controlled studies.Reference Ford4–Reference Karkos and Wilson7 Differences between studies are mainly related to patient selection and varying outcome criteria.
Our study showed that response to one-month omeprazole treatment was significantly associated with three factors: low pre-therapy anxiety score, high heartburn score and receiving omeprazole 20 mg twice daily (Table III). The last two factors significantly increased the odds (by approximately two and three times for positive response), while a high anxiety level decreased the odds. No significant differences were found for all other evaluated factors (Table II). The significance of anxiety may be explained by more anxious patients who are resistant in the assessment of their improvement as well as more distrustful of their LPR diagnosis.Reference Siupsinskiene, Adamonis and Toohill12 The significance of initial heartburn score as a relevant predictor for one-month empirical omeprazole treatment is not surprising, as heartburn is a classical symptom of gastroesophageal reflux disease which has a long history of good response to PPI treatment, as described elsewhere. A previous report advocating twice daily PPI treatment gave findings similar to this study. Twice daily PPI treatment is more effective than once daily treatment.Reference Bilgen, Ögüt, Kesimli-Dinç, Kirazili and Bor2, Reference Ford4, Reference Vaezi6, Reference Park, Hicks, Khandwala, Richter, Abelson and Milstein10, Reference Siupsinskiene and Adamonis19
Analysis of selected pre-therapy factors which may influence three-month empirical PPI treatment outcome showed no statistically significant associations. However, regarding duration of treatment, our study results decisively showed that three months of potent PPI treatment was successful for most carefully selected patients. After one month of ineffective treatment with PPIs, attention should be paid to the baseline videolaryngoscopic score. A higher score could indicate a further, positive response following three months of treatment (the maximum videolaryngoscopic score for further non-responders was low-seven points, of a possible 36 points). However, these data should be interpreted with caution, because of the small number of further non-responders (n = 6).
• Diagnostic criteria for patients with laryngopharyngeal reflux (LPR) are symptoms, videolaryngoscopic findings and positive response to empirical proton pump inhibitor (PPI) treatment
• This study investigated whether patients with suspected LPR benefited from one and three months of empirical PPI treatment, and whether any pre-therapy factors predicted the success of treatment
• Results suggest that three months' PPI treatment is warranted to confirm suspicion of LPR. Patients without psychological distress, with heartburn, and receiving omeprazole 20 mg twice daily were most likely to benefit from one month's empirical PPI treatment
In a few trials analysing predictors of response to PPI treatment, Williams et al. Reference Williams, Szczesniak, Maclean, Cole and Cook8 evaluated 20 LPR patients and found that neither baseline symptoms, endoscopic findings, laryngitis severity, oesophageal acid exposure (on pH testing) nor speaking voice measures predicted the outcome throughout three months of treatment. Park et al. Reference Park, Hicks, Khandwala, Richter, Abelson and Milstein10 found that pre-therapy interarytenoid mucosa and true vocal fold abnormalities were associated with a two-fold increase in symptom response after four months of PPI treatment. Garrigues et al.,Reference Garrigues, Gisbert, Bastida, Ortiz, Bau and Nos11 in a study of 91 patients with posterior laryngitis, found that response to omeprazole 20 mg twice daily was associated with a lower age (after three months) and a lower duration of laryngeal symptoms (after six months), but concluded that a consistent prediction of response could not be made. Our analysis showed no predictive value for age.
Our study could have been limited by including LPR diagnosis without pH monitoring, although we followed currently accepted recommendations.Reference Koufman, Aviv, Casiano and Shaw1, Reference Ford4 Because of the very small number of non-responders at three months of treatment, data on pre-therapy predictors of response should be interpreted with caution. Further investigation on this point is needed.
However, our study findings are in agreement with others, that is, that at least three months of empirical PPI treatment, of adequate dose, are required to confirm LPR diagnosis. Careful selection of patients is critical for positive treatment response. Our results suggest that patients without psychological distress and with heartburn, treated with twice daily PPI, are more likely to benefit from empirical PPI treatment more quickly, i.e. in one month.
Conclusion
Three months of twice daily PPI treatment is warranted in order to confirm the suspicion of LPR.
Three pre-therapy factors were revealed as significant predictors for one-month omeprazole treatment response: anxiety, heartburn score and medication dose. A high pre-therapy anxiety score decreased the odds ratio of a positive response 1.15-fold, while a high heartburn score and omeprazole 20 mg twice daily increased the odds by two- and threefold.
We found that patients' demographic parameters, laryngeal symptoms, videolaryngoscopic scores, endoscopic findings, vocal dysfunction degree, voice handicap index and general well-being assessment scores did not predict the one-month omeprazole treatment outcome.
No statistically significant association was found between any of the evaluated pre-therapy factors and the response to three-month omeprazole treatment.
