Abstract
OBJECTIVES: The aim of this study was to validate and determine the reliability of a Portuguese version of the Summated Xerostomia Inventory.
MATERIALS AND METHODS: This was a descriptive cross-sectional study which employed volunteers suffering from hyposalivation. Each patient responded to the Portuguese version of the questionnaire in the form of an interview. This was repeated again after 2 weeks.
A standard single question provided a validity check. Data were analysed to test its reliability, internal consistency, test-retest reliability using Cronbach’s alpha, total and inter-item correlation, and intra-class correlation. Floor and ceiling effects were assessed. Alpha was set at 0.05. Informed consent and local ethical committee clearance were obtained
RESULTS: A sample of 103 volunteers was recruited. Cronbach’s alpha was 0.84 and 0.87 for the first and second test administrations, respectively. The intraclass correlation coefficient value for the test-retest reliability was 0.93, and scores for the individual items ranged from 0.79 to 0.90. The correlation between the total score of the questionnaire and standard single dry-mouth question was 0.66, indicating a very good correlation.
CONCLUSION: Demonstrating excellent psychometric properties, the Summated Xerostomia Inventory (Portuguese Version), is a valid clinical tool for quantifying xerostomia in a dry mouth afflicted population.
Introduction
Oral cavity homeostasis depends greatly on proper salivary function (Ekström, Khosravani, Castagnola and Messana, 2012; Napeñas, Brennan and Fox, 2009). Individuals with salivary gland impairment and consequent diminished salivary flow experience a plethora of signs and symptoms such as greater susceptibility to intraoral infections, dental caries, periodontitis, mucositis, dysphagia, dysgeusia, dysosmia and difficulties in using dental prostheses, among others (Cassolato and Turnbull, 2003; Hopcraft and Tan, 2010; Napeñas et al., 2009; Villa, Connell and Abati, 2014). Thus, hyposalivation can lead to social, nutritional or behavioral changes and affect quality of life (Bergdahl and Bergdahl, 2000; Cassolato et al., 2003; Fox, 2007; Hopcraft and Tan, 2010; Sasportas et al., 2013; Shiboski, Hodgson, Ship and Schiødt, 2007; Turner and Ship, 2007).
Certain medications, head and neck radiation, and autoimmune diseases—such as Sjögren’s Syndrome (among others)—are the major causes of salivary gland hypofunction, characterized by a diminished salivary output (Cassolato et al., 2003; Fox, 2007; Lofgren, Wickstrom, Sonesson, Lagunas and Christersson, 2012; Shiboski et al., 2007; Thomson, 2005; Villa et al., 2014). Xerostomia is considered the subjective feeling of dry mouth (Villa et al., 2014) and may develop as the salivary output decreases, or it may arise de novo. The exact nature of the relationship between hyposalivation and xerostomia has yet to be determined (Hopcraft et al., 2010; Lofgren et al., 2012; Mignogna, Fedele, Russo, Muzio and Wolff, 2005; Napeñas et al., 2009; Turner et al, 2007). Both dimensions of salivary flow, unstimulated or stimulated, are regularly used as an objective indicator to evaluate dry mouth(Lofgren et al., 2012). Nevertheless, the level of xerostomia also needs to be assessed because it is, from a patient-centered perspective, an important outcome which can affect quality of life (Thomson, 2005). Moreover, the prevalence of xerostomia may be increasing as life expectancy extends (Lofgren et al., 2012; Thomson, 2005; Villa et al., 2014).
Being a subjective sensation, several questionnaires have been created for the measurement of the degree of xerostomia, and these range from a single item up to an 11-item questionnaire(da Mata et al., 2012; He, Wang and Li, 2013; Lee, Koh, Kwok and Park, 2016; Serrano et al., 2015; Thomson, Chalmers, Spencer and Williams, 1999). The single-item question has arisen first, but is limited (Hopcraft et al, 2010; Thomson, 2005). Multidimensional instruments have proved to be of greater utility since they have the capacity to place respondents on a continuum of xerostomia symptoms, thus producing more accurate evaluations of oral dryness (Thomson, 2007). The Xerostomia Inventory (XI) is an 11-item summated rating scale, which results in a single score representing the severity of dry mouth perception (Thomson, 2007; W M Thomson et al., 1999; van der Putten, Brand, Schols and de Baat, 2011). This questionnaire was developed in the 1990s and includes both the experiential and behavioral aspects of this condition. It has been shown to have acceptable psychometric features(da Mata et al., 2012; He et al., 2013; Lee et al., 2016; Serrano et al., 2015; Thomson et al., 2011; Thomson, 2005, 2007; Thomson, Williams, Spencer and Williams, 2000; van der Putten et al., 2011). Despite the continuing importance and utility of the single item approach, the XI has gained in popularity in salivary research in recent years (Garcia et al., 2009; Hopcraft and Tan, 2010; Randall et al., 2013; Sasportas et al., 2013; van der Putten et al., 2011; Villa et al., 2014).
First published in 2011, the shortened Xerostomia Inventory (SXI) is a short-form 5-item version of the original questionnaire(Thomson et al., 2011). This was done because some questions in the XI tapped into dimensions other than xerostomia alone, and also because a shorter version would be handier in the clinical environment. The scores for the 5 questions are summed, which results in a single score which represents, subjectively, the severity of xerostomia. The new questionnaire has been increasingly used clinically and in research with excellent results(da Mata et al., 2012; He et al., 2013; Lee et al., 2016; Serrano et al., 2015). A version of the XI-11 (XI-PL) in Portuguese was pre-validated in 2012(da Mata et al., 2012), but no Portuguese version of the SXI has been developed. Moreover, no systematic examination of the test-retest reliability of the SXI (or XI) has yet been conducted. Accordingly, the aim of this study was to extract, pre-test, validate and determine the reliability of a Portuguese version of the SXI (SXI-PL).
Materials and Methods
This was a cross-sectional study with the aim of developing a version of the SXI (SXI-PL) in Portuguese and assess its psychometric attributes like, its repeatability, validity and reliability. The original SXI is a questionnaire composed of 5 questions from which the respondent can choose from 3 available answers: “never” (scoring 1), “occasionally” (scoring 2) or “frequently” (scoring 3). The scores from the 5 questions are summed, with the result representing the degree of xerostomia the subject feels. In this study, a sample of patients that were concurrently recruited in the context of two different clinical trials conducted by our team were invited to participate.
The 5 questions to be included in the SXI-PL were picked from the original XI-PL(Thomson et al., 2011), which had been previously translated and validated in 2012(da Mata et al., 2012). They were the same items as those included in the SXI.
The resulting SXI-PL was analyzed by three specialists from the field of oral medicine. All agreed on maintaining the original phrasing. The final version of the SXI-PL is depicted in Table 1.
This study employed a sample of 103 volunteers suffering from hyposalivation. The inclusion criteria for the study were: (a) resting whole saliva flow rate < 0.2 ml/min; (b) stimulated whole saliva flow rate > 0.2 ml/min; (d) more than 18 years of age. Exclusion criteria were as follows: (a) wearer of complete dental prosthesis; (b) resting whole saliva flow > 0.1 ml/min; (c) those who were pregnant or lactating; and (d) non-speakers of Portuguese.
All eligible participants gave their written informed consent before study admission.
A full medical history was taken, and saliva collection was performed at the Portuguese Institute of Rheumatology and at several home-care facilities.
Unstimulated and mechanical-stimulated whole saliva flow rates were measured (5 min collection per sample) following established methods (da Mata et al., 2009) and expressed by ml/min +/- standard deviation (SD).
Each patient answered to the SXI-PL version of the questionnaire in the form of a standardized interview. Patients were told that the question asked had no definitive right answer and so were required to give the answer that immediately came to mind. Patients were asked if it was necessary for the interviewer to repeat himself or explain any aspect of the question that they could not understand before providing a response. This procedure was repeated with a 2-week interval, to enable the evaluation of the test-retest reliability of the SXI-PL.
Patients were also asked to respond “never” “occasionally” “frequently” or “always” to the single item: “How often does your mouth feel dry”. This was done to provide a concurrent validity check.
The study protocol was approved by the ethical committees of both the Faculty of Dental Medicine of the Lisbon University and the Portuguese Institute of Rheumatology. It was conducted in full compliance with the World Medical Association Declaration of Helsinki and its most recent amendments and always followed good clinical practice guidelines.
To analyze the data, the SPSS program (version 22.0; Chicago, IL, USA) was used. A patient was removed from the study if he failed to answer more than two questions. The dependent variable was the SXI-PL score, expressed as the summated score ± standard deviation (SD). Significance was set at α = 0.05.
Cronbach’s alpha was used to determine internal consistency of the questionnaire. This value was considered desirable and rated as good if it was at least 0.80 (Bland and Altman, 1997).
Despite the length of the 5 question questionnaire we also calculated the inter-item correlations to determine the possibility of inflation of the Cronbach alpha (Streiner, 2003). This value should be above 0.4 to be sufficiently reliable (Clark and Watson, 1995).
We also examined correlations of the individual questions with the summated score (item-total correlation), which should be above 0.3 (Ferketich, 1991; Serrano et al., 2015), and also if by removing a question the value of Cronbach’s alpha would be improved.
After the determined two week interval, the SXI-PL was completed by the patients in the same manner. The intra-class correlation coefficient (ICC) was determined to calculate test-retest reliability of every subscore and total overall score. The model used was two-way random with absolute agreement and 95% confidence intervals. ICC lower than 0.4 were considered to have low reliability while a range from 0.4 to 0.75 has good reliability. The optimal ICC values is higher than 0.75 (Fleiss, 1999).
The means of the total SXI-PL scores were plotted against the standard question response categories to assert concurrent validity. The correlation between the total XI-5-PL scores and the standard question responses were examined using Spearman’s Rho.
Floor and ceiling effects were a concern for the assessment of content validity. These were deemed to be influencing the questionnaire if more than 15% of the patients scored in the extremes of the overall summated score (McHorney and Tarlov, 1995)
Results
Salivary flows and age of participants are presented in Table 2. The study sample comprised 91 females (88.3%) and 12 males (11.7%). Mean total SXI-PL scores were 11.2 (SD, 2.9) and 11.6 (SD, 3.0) for first and second round answers, respectively.
Intra-class correlation coefficient (ICC) and item-total correlation (ITC) values are showed in Table 3. Positive correlations between all items in the inter-item correlation matrix were found. The values for the coefficients from the five questions ranged from 0.42 to 0.70. Average inter-item correlations were 0.52. For scale stability the values of the item total correlations ranged from 0.63 to 0.77. There was an equal contribution for scale dimensionality by each item in the scale. Cronbach alpha values for the 5 questions were 0.84 and 0.87 for the first and second test administrations respectively, and were lower if any items were removed, suggesting that all 5 questions contributed positively to the questionnaire´s internal consistency. The questionnaire showed excellent reliability with the intraclass correlation coefficient value for the test-retest reliability of the SXI-PL score being 0.93; and ICC scores for the individual items ranged from 0.79 to 0.90. A modest ceiling effect was found, with 18% of participants scoring the maximum value. Finally, there was a strong positive correlation (Spearman’s Rho = 0.66) between the standard item response and the SXI total score. Moreover, when plotting mean SXI scores by standard question responses, a statistically significant gradient across the categories of the standard question was observed (Figure 1).
Discussion
This study set out to validate a Portuguese version of the SXI (SXI-PL). The findings suggest that the SXI-PL is a reliable and valid form of measuring xerostomia, in keeping with its parent English version.
The study has several weaknesses. Namely, the sample could have been more diverse and greater in size, since only aged patients on medication and patients with Sjögren´s Syndrome (SS) were included. This is likely to have affected the external validity of the study and hence, the generalization of the findings. However, medication and SS are major causes of xerostomia, and the findings suggest a good performance of the scale. Moreover, this was a pretest study and a more comprehensive study should be designed to confirm these findings.
A major strength of this study was the double administration of the questionnaire with a separate time interval, thus enabling the first assessment of the test-retest reliability of the SXI. Intra-class correlation scores ranged from 79% to 93% with a two-week interval indicating excellent test-retest reliability for the SXI-PL. This results were similar or even elevated when compared to other studies.(Lee et al., 2016). This is an important finding since the test-retest reliability of any questionnaire is a critical characteristic. Moreover, the mean Cronbach alpha value of 0.85 obtained in this study suggests that the 5 questions in the XI-5-PL are measuring the same construct and have good internal consistency. Similar findings have been obtained in other studies(da Mata et al., 2012; Gkavela, Kossioni, Lyrakos, Karkazis, and Volikas, 2015; He et al., 2013; Lee et al., 2016; Serrano et al., 2015; Thomson et al., 2011). Positive correlations between all items were found, with a mean inter-item correlation of 0.52; which is desirable according to Clark and Watson (Clark and Watson, 1995), who consider values of 0.40 to 0.50 to be required for scales tapping narrow characteristics as xerostomia. Similar values were found in other studies(da Mata et al., 2012; Lee et al., 2016; Serrano et al., 2015) although higher than in others (Gkavela et al., 2015).
Strong correlations (0.63 to 0.77) were also found when comparing an item and the rest of the scale (item-total correlation), all well above the recommended threshold (0.3) for including an item in a scale; these finding are consistent with those of other studies (da Mata et al., 2012; Lee et al., 2016; Serrano et al., 2015). Additionally, removing one or more items resulted in lower Cronbach alpha values, reflecting the contribution of each item to the overall internal consistency of the inventory. Finally, there was also a significant positive and strong correlation between the single item question and total score, fulfilling the criteria for independent validation as proposed in previous studies (da Mata et al., 2012).
The Portuguese language is spoken by more than two hundred million people worldwide. The validation of a simpler and reliable xerostomia questionnaire should therefore be considered as a valuable tool for clinical use. These findings are more important than ever because of increasing population longevity and the associated increase in numbers of older people with multimorbidity and polypharmacy. Monitoring and treating their dry mouth symptoms will be an important aspect of their health care.
Within the limitations of this study we can conclude that the XI-5-PL has excellent psychometric properties and can be used successfully has a tool to measure xerostomia of patients with hyposalivation.