(Atkinson et al., 1989', Johnson et al., op. cit.; Loesche ... In a notable exception, Narhi, er al., (op. cit.) ..... does indeed show this, but the degree of overlap (as.
Community Dental Health (1999) 16, 12-17
O BASCD 1999
Received 28 January, 1998; accepted8 April, 1998
The XerostomiaInventorv: a multi-item approachto measuringdry mouth W. Murray Thomsonr,JaneM. Chalmers2,A. John Spencer2and SheilaM. Williams3 lDepartment of Oral Health, The University of Otago, Dunedin, New Zealand 'zAIHW Dental Statistics and Research (Jnit, University of Adelaide, South Australia 3Departmentof Preventive and Social Medicine, University of Otago, Dunedin, New Zealancl
Objective To develop a valid multi-item method of measuring the symptoms of xerostomia which includes the wide range of xerostomia symptoms in a single quantitative measure. Design A combination of qualitative and quantitative approaches. Setting A cohort study in South Australia. Participants Older people aged 65 years or more who were taking part in the South Australian Dental Longitudinal Study. Measures Xerostomia symptoms were evaluated using a multi-item inventory format and, for comparison pulposes, a standard single dry-mouth question. Resting whole-salivary flow rate was estimated using the 'spit' method. Resulls Xerostomia and flow-rate data were available for 636 individuals. Factor analysis revealed the presence of a discrete xerostomia dimension, representedby l1 items whose responseswere summated to give a single Xerostomia Inventory (XI) scale score. This had a very low correlation with resting flow rate but a much stronger, positive correlation with the standard dry-mouth question responses.Conclusinns The XI shows adequatecontent and concurrent validity, and appearsto be a promising advance on previous approachesto xerostomia symptomatology although further testing is required. Key words: dry mouth, measuremen| sialometry, xerostomia
Introduction Xerostomia is the subjective feeling of dry mouth. Estimates of its prevalence in older populations range from 10 to 38 per cent (Osterberget al., 1992; Thomson et a1.,1993;Locker, 1995).Salivary gland hypofunction (SGH) is a measurablereduction in salivary output; estimatesof SGH prevalenceamong older adults are more difficult to compare because different approaches and definitions have been used.Johnsonet al. (1984) reported that 44 per cent of an institutionalised older population had a stimulated parotid flow rate of 0.2 mVmin or less. Osterberg et al. (1984) reported that 32 per cent of a representativesample of Swedes aged 70 or more had a resting whole-salivary flow rate below 0.1 mVmin, while Fure and Zickert (1990) reported a prevalence rate of22 per cent for the same flow rate among a representative sampleof 55-, 65- and 75-year-oldSwedes.It is apparent from the prevalence estimates for xerostomia and SGH that substantial numbers of older people are affected. Evidence suggests that older people's salivary systemsare susceptibleto exogenousfactors that may act to reduce their secretory capacity, such as medications (Atkinson et al., 1989',Johnsonet al., op. cit.; Loesche et al., 1995',Narhi er al., 1992; Osterberg et al., 19841' Persson et al., l99l; Thomson et al., op. cit.), some chronic inflammatory conditions (Sreebny and Valdini, 1989), and radiation such as that received during radiotherapy for cancer of the head or neck (Dreizen et al., 1977). Becatse these exogenous factors are common
among older adults, the prevalence of both xerostomia and SGH is higher among older people (Baum 1989), and hasbeen shownto increaseover time (Locker, 1995). The measurement of xerostomia is problematic. In contrast to SGH, which can be objectively evaluatedby using sialometry(Navazesh,1993),xerostomiais comprised of a setof symptoms,and thereforecan be assessed only by directly questioningindividuals (Fox et a1.,1987). Despite the many studies that have investigated xerostomia, no insffumenthas been developedwhich clearly and distinctly represents the condition so that its measurement can be undertakento allow comparisonsacrossgroups. In some reports, the method of assessingxerostomia has not been specified, except for reference to patient-reported changes in symptoms (Greenspan and Daniels, 1987; Fox et al., 1986').In other reports, a single-item questionnaire(Neverlien, 1994)has been used (Fox er al., 1987; Gilbert et al., 19931,Johnson et al., op. cit.; Locker, 1993, 19951,Narhi, 1994; Osterberg er al ., 1992;'Thomson et al., op. cit;). There is no way of determining exactly what is being measured with such approaches.The researchermay assumethat a single item representsa range of severity of xerostomia symptoms, but in reality it may be measuring a completely different dimension because interpretation of the single question may differ amongrespondents.Moreover, such single-item approachesto objectively measuringxerostomiahave been limited to dry mouth only, without exploring the wider constellation of symptoms which appear to form the xerostomia experience.
Correspondenceto: W. Murray Thomson, Department of Oral Health, The University of Otago, PO Box 647, Dunedin, New Zealand
In a notableexception,Narhi, er al., (op. cit.) took a broader approachand employed a seriesof items ranging from the experience of a continuously dry mouth, to difficulty in speaking and in swallowing. Less frequently-explored symptoms such as oral burning or itching sensationswere also investigated,and were found to be more prevalent in individuals with dry mouth, more of whom also repofied taste impairment and difficulties in eating dry foods. This more inclusive approach was a useful development, but each item was still analysed in isolation rather than being aggregatedinto an inventory. A more integrated approach would provide a continuous score which would allow more precise measurementof the severity of symptoms, and reduce the potential for misclassificationwhich occurswith categoricalapproaches to defining the condition. Valid expressionof the total concept of an experience such as xerostomia requires a derived measure which is a combination of many different, fundamental and directly observed measures (Kaplan et al., 1976). In measuring xerostomia, it is assumed to be a complex phenomenon which can be represented by a latent variable for which direct measurement is not possible, and which must therefore be estimatedby making observations of a set of relevant indicator variables. Clinical depressionoffers a ready analogy: to label an individual 'clinically as being depressed' means that he or she exhibits a set of symptoms which commonly occur in conjunction with one another; however, no single indicator exists for clinical depression (Bartholomew, 1996). Out of these considerations (and the plethora of different items which have been used to date) arose the
notion that this approach may be useful for measuring xerostomia. The purpose of this paper is to describe the development and preliminary testing of the Xerostomia Inventory, a multi-item instrument for measuring xerostomia symptoms which enablesan estimate of their severity to be made on a continuous scale. Method The development of the XI involved a combination of qualitative and quantitative techniques. A literature search revealed a number of items which had been developed and used by other workers using single-item inventory approaches(Table 1), and this enableda framework to be developed for semi-structuredinterviews which were undertaken with a convenience sample of four diagnosed long-term sufferers of xerostomia who were patients in the high caries clinic at the Adelaide Dental Hospital. It is not known how representativethose individuals were of xerostomia sufferers in general, but they appearedto be typical ofpatients seenin that clinic. Responseswere recordedin longhand, and content analysis was used to identify dominant themes which were then either developed into new XI items - using the interviewees' own words where possible - or used to confirm and/or modify those which had been obtained from the literature. This process resulted in 19 separate items, and ensured: (1) that those which were used reflected many manifestations of the xerostomia experience; (2) that their most appropriate wording was determined; and, (3) that they were grounded in the
Table L, Previousapproaches to investigatingxerostomia Items used specified in report
Workers
Does your mouth feel distinctly dry? Do you sip liquids to aid in swallowing dry foods? Does your mouth feel dry when eating a meal? Do you have difficulties swallowing any foods? Does the amount of saliva in your mouth seem to be too little, too much, or you don't notice it? Do you feel dryness in the mouth at any time? Do you have mouth dryness? Is your mouth sometimes dry? How often does your mouth feel dry? During the last four weeks, have you had any of the following....dryness of mouth? Does your mouth feel dry? Does your mouth usually feel dry?
Osterberg et al., 1984 Fox et al.,198'7 Fox et a1.,1987 Fox et a1.,1987 Fox et a1.,1987 Fure & Zickert;1990 Osterberg et al., 1992 Gilbelt e/ al., 1993 Thomson et al., 1993 Locker, 1993 Narhi er al.,1992 Nederfors et al., 1997
Exact wording of items used not specffied in report but indication given of approach which was used Difficulty in eating dry foods Difficulty in speaking Difficulty in swallowing Taste impairment Dry lips Burning sensation in oral mucosa Burning sensation in tongue Itching sensation in oral mucosa Itching sensation in tongue Mouth breathing Dry throat Dry nose Dry skin Drv eves
Narhi Narhi Narhi Narhi Narhi Narhi Narhi Narhi Narhi Narhi Narhi Narhi Narhi Narhi
1994 1994 1994 1994 1994 1994 1994 1994 1994 1994 1994 1994 1994 1994
t-l
experiencesof xerostomia sufferers. The 19 items were then assembledinto inventory fotmat for testing in the five-year data collection phase of the South Australian Dental Longitudinal Study (SADLS). Responseoptions were presented in a large font for ease of reading by older adults, with each response item having the same 'Hardly 'Occasionally', five options: 'Never', ever', 'Frequently',or'Always'. The SADLS sampling strategyand data collection has been describedpreviously (Slade and Spencer, 1994), but a brief description follows. For the baseline study in 1991, a stratified random sample of non-institutionalised persons aged 60+ years was selected from the compulsory State Electoral Database of non-institutionalised Australian citizens (aged 18 or more). The sampling method defined 24 strata, 18 in the Adelaide Statistical Division, and 6 within the Mt Gambier City and District Council. The former comprised three age groups (60-64, 65*14 and 75+ years), two genders and three locality groups (which were basedon their distance from a public or school dental clinic), while the latter was comprised of the same three age groups and two genders. A different selection probability was used for each stratum in order to draw a simple random sample of participants. Dentate people (those who had one or more natural teeth) were over-sampled by excluding a percentageof edentulouspeople which ranged from 100 per cent in Mt Gambier to 50 per cent among the Adelaide residents aged 60-64 years. The baseline and two-year data collections took place in l99l and 1993 respectively. At five years (1996), the participants were again examined and interviewed, with computer-assisted telephone interviews being used to collect household and personal information just prior to the clinical examination. During the telephone interview stage of the study, responsesto a standardsingle xerostomiaquestion ("How often does your mouth feel dry?" Responseoptions: ' N e v e r ' , ' O c c a s i o n a l l y ' , ' F r e q u e n t l y ' , ' A l w a y s 'w ) ere collected. The Xerostomia Inventory was sent as a postal questionnaire to all parlicipants who had agreed to a clinical examination, and they were instructed to either bring the completed questionnairesto the clinical examination or return them by post. There was no systematic follow-up of non-responders to the postal questionnaire, and so information on the reasons for refusal was not obtained. Resting whole saliva was collected at the clinical 'spit' examination appointment using the method (Navazesh and Christensen, 1982). Each participant had been instructed to refrain from food, beverages and smoking for the 60 minutes prior to collection. Some five minutes before saliva collection, participants were instructed to rinse the mouth out with plain water and then to sit quietly while administrative procedures were completed. Immediately prior to saliva collection, each was asked to clear the mouth by swallowing, and then to actively spit saliva into a pre-weighed plastic collection tube over the next four minutes. At the end of that time, a beeper sounded and the participant was asked to spit any remaining saliva into the tube, which was then sealed and placed in a cool-storagebin. The collection time was recorded. The tubes were weished later at the Adelaide
14
Dental Hospital. Resting saliva flow (in ml/min) was computedas the weight of saliva collected(assuming1g = lnrl) divided by the collection time in minutes. Data were entered into an SPSS data file and analysed using SPSS. Univariate statistics were computed. No prior hypotheses about the factor structure of the data were available, so exploratory factor analysis was undertaken using Maximum Likelihood methods. Cronbach's o was computed to confirm the internal consistencyof responsesto the xerostomia items prior to the collapsing of their responsesinto a single continuous factor scale. Bivariate analysesused analysis of variance. Results Of the 939 people (56.9 per cent of baseline) who remainedin the study at five years,483 (51.4 per cent) were male and 456 (48.6 per cent) female. In comparison with the baseline characteristicsof those who were lost to follow-up, the individuals who remained were younger, had fewer missing teeth, fewer chronic medical conditions, and were more likely to be regular users of dental services. The ages of the remaining study members ranged from 65 to 100, with a mean age of 75 years (SD, 7 years). Xerostomia questionnaireswere mailed to the 708 (75.4 per cent) who had a dental examination appointment. The questionnaires were completed and returned by 649 (91.7 per cent) of those individuals, of whom 201 (31.0 per cent) were from Mt Gambier and 448 (69.0 per cent) were Adelaide residents. Where there were difficulties in getting co-operation from participants, priority was given to the dental examination rather than the saliva collection; consequently, saliva samples were collected from 676 (95.5 per cent) of those examined. Both xerostomia inventory questionnaireand salivary flow-rate data were available for 636 individuals (89.8 per cent), and 662 (93.5 per cent) supplied saliva samplesand answeredthe standard question. Responsesto individual items are shown in Table 2. The factor analysis revealed three eigen values greater than 1 (5.06, 1.85 and 1.26 respectively),and solutions with 1, 2 and 3 factors were considered. Inspection of the loadings after varimax rotation - as well as examination of the scree test - suggested that two factors provided the best solution. The factor loadings which show how each item correlated with the underlying latent variable are shown in Table 3. Two scales were constructedby summing the responsesto the items which loaded on each factor. The correlations between the unweighted factor scales and those constructed with optimal weighting were 0.93 for the first scale and 0-95 for the second,indicating that the unweighted summated scales representedthe underlying factors well. Coefficient o was 0.84 for the first scale, and 0.80 for the second scale. The means for each scale are shown in Table 4. The Factor 1 scale was designated the Xerostomia Inventory (XI) score, while the Factor 2 scale was named the Burning Mouth Syndrome (BMS) score. The correlation between the two scales was positive and significant (r = 0.40; P
