There is evidence of a variation in salivary flow rate over a 10-month interval in children. The relation between salivary flow rate and salivary constituents suggests that this variation might have a secondary effect on the measures of inorganic ions and buffering capacity. A fluctuation independent of salivary flow rate might also be present. For example, in unstimulated submandibular saliva, circadian variation in calcium and phosphate is observed. Yet, even when salivary flow rate is controlled, there is evidence that calcium varies throughout the day in both parotid and submandibular saliva.

Variation in salivary constituents over longer intervals of time has also been reported; Shannon and Segreto (1968)observed that calcium varied haphazardly from day to day and from week to week in stimulated parotid saliva. Phosphate increased throughout the week and also showed a week to week fluctuation. However, when salivary flow rate was controlled, no significant difference in the amounts of inorganic ions was observed from day to day in parotid saliva.

Variation of salivary constituents over time could simply reflect other external influences. Hormonal factors might influence salivary constituents over longer intervals. For instance, salivary ionic calcium is higher in the mid-menstrual cycle than pre- and postmenstrually.

If a variation of salivary constituents does exist, comparison of single salivary measurements collected under the same conditions, even in a cohort, may not be appropriate. We have now examined the variation of salivary calcium, phosphate and buffering capacity over a 10-month interval. The study was carried out in North Wales where a group of 43 children, 11–12 years of age, provided a mid-afternoon saliva sample once a month between September 1990 and June 1991 inclusive. The dates of salivary collection were separated by between 28 and 33 days except on one occasion when collections were 23 days apart. The sample was collected 1 hr after eating and the dates of commencement of menstruation of female participants (n=19) were recorded each month. Details of the timetable and method of collection of the unstimulated saliva sample are as described by Kavanagh et al. (1998).

Buffering capacity was assessed by the Dentobuff® strip method (Orion Diagnostica, Espoo, Finland) immediately after the collection of each saliva sample. Results were categorized as low (endpoint pH<4.6), medium (endpoint pH between 4.6 and 5.5) or high (endpoint pH between 5.6 and 7.0) according to the colour chart provided.

The remainder of the saliva samples were stored at ?18°C and then, after thawing, analysed in the Department of Chemistry, University College Cork, Republic of Ireland. Total calcium was measured by atomic absorption spectrophotometry in a Perkin–Elmer Model 2380. Spectrophotometry was used to measure inorganic phosphorus. The coefficients of variation of reproducibility were <8% for calcium and <5% for phosphorus.

Eighteen individuals attended at every occasion throughout the study. Although saliva was collected from each individual who attended, not all of the samples were of an adequate volume to analyse buffering capacity, calcium and phosphate. Where there were missing values for buffering capacity, these were substituted by the series modal value and with regard to calcium and phosphate a series mean was used. Intra-individual variation in buffering capacity and inorganic ions was assessed by Friedman two-way ANOVA and repeated-measures ANOVA. When such variation was statistically significant, multiple pair-wise comparisons were made using the Bonferroni modification of the significance level of a t-test to identify if the variability was restricted to a few comparisons. Multiple-comparison tests were only used for continuous data. The analysis was done with SPSS software (SPSS Inc., Chicago, Illinois, U.S.A.).

The percentage of children recording high readings from Dentobuff® strips was never less than 27%. Less than 17% of individuals were in the low Dentobuff® class except in October and June when over 20% were in this category. Intra-individual variation in this total group was statistically significant (Friedman two-way ANOVA, p<0.001). There was no statistical evidence that this was related to sex.

The numbers of samples analysed for calcium ranged from 28 in February to a maximum of 37 during the months of March, April, May and June. The group average ranged from 1.38 mmol/l in December to 1.83 mmol/l in January. Intra-individual variation throughout the investigation was statistically significant (repeated-measures ANOVA, p<0.001). When segregated according to sex, no statistically significant difference was observed.

Multiple pair-wise comparisons revealed that the paired comparisons November vs December, December vs January and November vs February were statistically significant (paired t-tests Bonferroni method, p<0.001). Yet, when these 3 months were excluded individually from repeated-measures ANOVA, intra-individual variation remained statistically significant (repeated-measures ANOVA, p<0.05). When both November and February were removed from analysis, intra-individual variation remained significant (repeated-measures ANOVA, p=0.002). Only when the two months of the paired comparisons, November vs December (repeated-measures ANOVA, p=0.086) and December vs January (repeated-measures ANOVA, p=0.081), were excluded was the intra-individual variation no longer statistically significant.

The number of samples available for phosphate analysis varied between 27 collected during the month of February to a maximum of 37 in March and April. The group’s means were highest in September and April, 5.41 mmol/l and 5.40 mmol/l, respectively. The lowest group average was in December, 4.96 mmol/l. A repeated-measures ANOVA did not detect a significant intra-individual variation for the duration of the study (p=0.14).

Intra-individual variation of salivary buffering capacity and calcium was here observed over a 10-month interval. While intra-individual biological variation is not unusual, predictive tests of disease based on salivary composition assume that these variables are reproducible on the majority of occasions within an individual. This variation may reflect salivary flow-rate variation, primary fluctuation of salivary constituents and external influences such as hormonal influences. The methods used to analyse the various constituents could also have contributed to the variability observed.

There was some evidence from multiple-comparison tests that the contrast between individuals’ salivary calcium in December compared with November and January was within the time when the majority of intra-individual variation occurred. This echoed the group means, where December had the lowest mean calcium (1.38 mmol/l; n=35) recorded and the adjacent months, November and January, corresponded to the two highest means (1.82 mmol/l; n=35 and 1.83 mmol/l; n=35). When comparisons of December with its neighbouring months were excluded from the repeated-measures ANOVA, intra-individual variation was no longer statistically significant. However, this loss of significance could have been confounded by a reduced sample size. November vs February pair-wise comparisons were also statistically significant in multiple-comparison tests, yet series of measures for both these months were excluded from repeated-measures ANOVA, statistical significance remained (p=0.002). February had the least number of samples available for paired intra-individual analysis and so its elimination would have had a smaller effect on sample size. Furthermore, the maintenance of statistical significance despite the removal of November, December and February individually from repeated-measures ANOVA suggests that intra-individual variation cannot unequivocally be assumed to be wholly restricted to a few months.