With these, participants respond to each item in a list, usually with a ‘yes’/‘no’ response format. At the analysis stage, the number of positive responses is counted and used as an index score, either as a simple count or after recoding of that count into ordinal categories. Such methods have been used by a number of workers (Table 7.2). For example, [46] used a list of seven questions (sourced from the literature) and a simple ‘yes’/‘no’ response format to assign nursing home residents to the following three groups: no xerostomia (0 positive responses), mild xerostomia (1–2) or marked xerostomia (3–7). The battery was found to have acceptable internal consistency reliability and was able to distinguish respondents with poorer OHRQoL from those with better OHRQoL, but there was no comparison with a global xerostomia item, meaning that no judgement of its validity as a xerostomia scale could be made.

Another such approach is the Challacombe scale [23], which was designed for clinical use as an objective score for oral dryness, based entirely upon the clinical observations of the examining clinician. Strictly speaking, it should not be included here because it purports to measure salivary gland hypofunction and has no subjective aspect; the individual being assessed is not asked about his/her symptoms. However, it is included here for completeness, because it is likely to turn up on any literature search for xerostomia indices. That particular scale appears to be a checklist which has arisen from experienced clinicians’ observations over many years. Some clinical validity has been demonstrated, but its utility remains unclear because of the wide range of entities which comprise it. Included in the 10-item checklist are observations on instruments adhering to mucosal surfaces, saliva frothiness, whether saliva pools in the floor of the mouth, tongue appearance, gingival architecture, mucosal appearance, palatal debris and recent experience of root surface caries. The provenance of that combination was not directly specified, but it is likely to have arisen from direct clinical observations by the experienced clinicians who were involved in its development. As with the battery used by Locker, the index score is a simple count of the number of signs observed and can range from 0 to 10. Its initial validation was undertaken against measurements of salivary flow rate and oral mucosal wetness in convenience samples of Sjögren syndrome patients and a comparison group (also a convenience sample) of university staff and rest home residents. The scale showed promising validity, but it should be further examined in population-based samples and against subjective measures of dry mouth. Until that has occurred, its utility remains unclear.

Battery-type approaches can yield meaningful scores and useful data for exploring the determinants of xerostomia. However, a battery of items suffers from the problem of being really just a ‘present/absent’ checklist of items or issues which may or may not relate to an underlying construct (such as the experience of dry mouth).

A modification of the battery-type approach was that used by Pai et al. [26], who used a battery of eight items, with each scored using a visual analogue scale (VAS). A VAS employs a line upon which the respondent places a mark to indicate the point which represents their position between the two extremes. That particular instrument had eight xerostomia-related items, each of which had a VAS response format (ranging from 0 to 100 mm, with 100 being the worst score). Other than the use of a VAS (rather than a Likert scale, which has ordered categories) for the responses, this is essentially just a variation on the abovementioned two. It is, if anything, more restricted in its use because the individual item scores are not used together. Gerdin et al. [12] used the same VAS instrument to measure xerostomia among Swedish nursing home residents. The point of difference with this particular study was that the individual item VAS scores were then summed to give an overall score (which could range from 0 to 800 scale points). Those scores were then used to allocate respondents to one of two symptom severity categories (‘no or weak dry mouth symptoms’ or ‘symptoms’) using a cut-off value determined from responses to the item ‘Does your mouth feel dry?’ first used by Fox et al. [10]. The resultant scale scores showed strong correlations with other dry mouth self-reports but no association at all with salivary flow rate. Summing the responses from a VAS-type response format is uncommon but has been described as acceptable [7].

A problem with the scales described above is that their constituent items were assembled somewhat arbitrarily, from the literature, clinical experience or a combination of the two. They may have a degree of empirical validity (as evident in score gradients and associations in the hypothesised direction when they are tested clinically), but there may still be a considerable amount of error and unexplained variance which is due to the inclusion of less relevant items (or indeed the omission of others which might have been more useful). This is where summated rating scales have distinct advantages.

A summated rating scale is a multi-item scale which purports to measure an underlying construct (known as a ‘latent variable’). It is essentially a more refined, focused development of the item battery, with the distinction that the items have all been shown to be correlated with the latent variable and that the number of items in the scale has been demonstrated to be adequate. The idea behind using such a scale is to be able to place respondents on a continuum which represents the range of experience of the entity being measured (from the minimum to the maximum). Everyone should be able to be placed somewhere on that continuum.

Least possible ———————————————————— Most possible

The advantage of using such a scale score is that more subtle differences in health states can be explored; the data are used as a continuous variable rather than as a binary or ordinal variable.

The developmental sequence for such scales involves a fairly standard series of steps: conceptual development, in terms of the underlying theory and models; generation of an item pool, which is a comprehensive list of all issues which are relevant to the domain which it is planned to measure (from the literature, clinical experience, interviews with sufferers, and so on); item pool reduction and psychometric testing (using methods such as exploratory factor analysis to determine whether the items do actually relate to the underlying latent construct); and field testing and validation. Subsequent steps (which are generally useful) may be the development of short-form versions and validation/adaptation for use in cultures other than that in which the measure was first developed.

The Xerostomia Inventory (XI) is such a summated rating scale [31]. It is an 11-item scale (Table 7.3) which was developed during the mid-1990s as part of an investigation into the question of whether medications which cause dry mouth (xerogenic drugs) among older people are associated with great caries experience. At the time, it was realised that there was no satisfactory way of allocating participants to a continuum of symptom experience; that is, to give them a continuous score which could capture more subtle interpersonal differences in dry mouth symptoms than a global item with four or five response categories. It was hoped that being able to obtain such a continuous score would then enable such a score’s use in multivariate models of caries incidence and increment (the final, somewhat surprising outcome of the study was reported in [34, 35]).