The effect of mobile personalised texting versus non-personalised texting on the caries risk of underprivileged adults: a randomised control trial

Background In the Republic of Ireland (RoI), fluoridation has been effective and efficient for caries prevention at population level, regardless of income status; however, at individual level it still has limitations. This study aimed to compare personalised versus non-personalised text messaging on ‘chance of avoiding new cavities’ with the Cariogram, a computer-based caries risk assessment (CRA) model, in an economically disadvantaged adult population in the RoI. Methods The intervention was via a CRA summary letter plus 24 weekly personalised mobile-phone short text messages (text messages) based on the individual’s CRA, compared with a non-personalised approach via a non-personalised letter and a predetermined, fixed set of 24 weekly text messages. The study was designed as a two-arm parallel-group, single-blinded (assessor), randomised controlled study in County Cork, RoI. The primary outcome was a comparison of ‘chance of avoiding new cavities’ calculated by the Cariogram with clinical examination, interview, CRT® (Ivoclar Vivadent, Liechtenstein) and three-day food diary between the two groups at follow-up. We combined stratified randomisation with blocked randomisation for 171 participants who completed baseline. Of them, 111 completed follow-up and were analysed (56 and 55 from the personalised and non-personalised groups, respectively). Due to protocol violations, both intent-to-treat (ITT) and per-protocol analyses were conducted. Results The ITT analysis did not show a personalised intervention effect on ‘chance of avoiding new cavities’. Of the secondary outcome measures, only the stimulated saliva flow factor showed a personalised intervention effect, p = 0.036, OR = 0.3 (95% CI = 0.1, 0.9). The per-protocol analysis with 21 personalised and 33 non-personalised participants within two-message deviations showed no significant effect on ‘chance of avoiding new cavities’. Conclusions The null hypothesis in regard to the primary outcome for both ITT and per-protocol analyses was not rejected; however, as the minimal clinically important difference was included in the 95% CI for the per-protocol analysis, replication studies will be worth conducting to explore the potential of mobile devices for individual caries risk reduction. Trial registration University Hospital Medical Information Network Clinical Trials Registry (UMIN000027253) on 10 May 2017. The study was retrospectively registered. Electronic supplementary material The online version of this article (10.1186/s12903-019-0729-1) contains supplementary material, which is available to authorized users.

The 'Caries experience' parameter is a relative score with reference to local epidemiological data [2]. The current study used the latest available Irish adult data [1] as its reference. Table presents the cut-off scores of the D3cMFS and D3vcMFS values from the reference data for the 25 th and 75 th percentiles by age group (16-24, 35-44, 65+). Cut-off scores for the 25 th and 75 th percentiles at ages 20-, 40-and 70-years were plotted. 1 D3vcMFS includes non-cavitated where there was a definite shadow under the enamel, indicating the presence of dental caries that had progressed to dentine, but cavitation had not yet occurred. 2 With the assumption that the D3cMFS and D3vcMFS values increase in a straight line according to age, straight lines between the scores at 20 and 40 years of age and between 40 and 70 years of age for the 25 th and 75 th percentiles were drawn. If a medical-card patient's D3cMFS index fell below the 25 th percentile line, the a medical-card patient was scored as Score 1 (better than normal). If a medical-card patient's D3cMFS index fell above the 75 th percentile line, the medical-card patient was scored as Score 3 (worse than normal). If a medical-card patient's D3cMFS index lay between the 25 th and 75 th percentile lines, the medical-card patient was scored as Score 2 (normal for age group). If the D3cMFS index of the medical-card patient fell on one of the lines, the worse score was taken. The D3vcMFS index was referenced in the same manner. If the medical-card patient had > 2 active root caries lesions or > 2 enamel lesions, the medical-card patient was given Score 3. Score 0 means that the patient was caries free and had no restorations.

'Related diseases'
General diseases or conditions which can directly or indirectly influence the caries process, were listed as follows [2]: • any autoimmune disease (e.g. Sjögren's syndrome) • diabetes mellitus • anorexia nervosa • visually impaired • any manual dexterity which might cause them difficulties with cleaning their teeth properly • any disease which requires continuous medication that affect their saliva secretion • any condition requiring radiation to the head-neck region. Score 0 was given for patients with none of the general diseases above (no disease). Score 1 was given if there any of the general diseases above was present (mild degree). The Cariogram Manual stated Score 2 should be given if the patient was bedridden or may need continuous medication (severe degree, long-lasting). Because medical-card patients taking part in the current study were not bedridden and the definition of 'long-lasting' was unclear, Score 2 was considered as not applicable.
'Diet contents' Salivary lactobacillus count with CRT® saliva test (Ivoclar Vivadent, Liechtenstein) was used as an indicator of the 'diet contents' parameter [2]. Although retention areas, open cavities or bad fillings could contribute to a high LB score [2], these conditions were not considered in the current study. This parameter was scored using the manufacture's chart. Scores 0 and 1 were < 10 5 colony forming units (CFU)/ml saliva. Scores 2 and 3 were ≥ 10 5 CFU/ml saliva.
The distinction between Scores 0 and 1 and between Scores 2 and 3 were made according to the manufacture's chart. The interpretation of scores was as follows: starches and sucralose were included in the basic count of fermentable carbohydrates. Although strictly speaking vegetables have natural sugars, they were not counted as part of fermentable carbohydrate intake because some of the educational text messages encouraged eating vegetables rather than sugary foods as snacks.
When the medical-card patient did not write their bedtime and the medical-card patient had fermentable carbohydrates at 10 pm or later, one intake count was added. When the medicalcard patient wrote their bedtime and had fermentable carbohydrates within one hour before bedtime, one intake count was also added. The scores for this parameter are as follows: Score 0: 0-3.0 times/day (very low diet intake frequency) The decision to apply these adjustments was made on 16 April 2015, just before risk assessment and randomisation were performed for the first patient.

'Mutans streptococci'
Salivary mutans streptococci (MS) count with CRT® was scored using the manufacturer's chart which says Scores 0 and 1 were < 10 5 CFU/ml saliva and Scores 2 and 3 were ≥ 10 5 CFU/ml saliva. Note that the Cariogram was originally designed to use Dentocult® saliva test kits [2]. According to the CRT® instruction, CRT® bacteria correlates with the Dentocult® system; however, CRT® MS reacts more sensitively and is able to detect even low bacterial count. Both tests have a model chart with four pictures assessing the density of CFU/ml saliva.
The distribution of MS in the current study showed much lower risk than shown by other studies [3][4][5] and clinical data from two Japanese dental practices using Dentocult SM® (Oral Care Inc., Tokyo) (Additional file Table 2), although the current study population was expected to be economically disadvantaged (i.e. a high-risk group). Therefore, Score 0 was rounded up to Score 1 and Score 2 was rounded up to Score 3 in the current study.
6 Additional file Table 2   'Saliva secretion' parameter The volume of stimulated saliva collected over five minutes was collected using CRT® saliva tests. Unstimulated saliva was not measured in the current study. In the dental practice with a normal appointment between 9 am and 5 pm, the medical-card patient sat upright and stimulated salivation by chewing a paraffin pellet for five minutes. The saliva was drooled into a disposable graduated test tube through a disposable funnel during the collection period. The dentist measured the volume of the saliva in the test tube from the lowest point on the meniscus, the measurement did not include the foam, if any. The four-level scoring system is as follows: Score 0: ≥ 1.1 ml/minute (normal saliva secretion) Score 1: < 1.1, ≥ 0.9 ml/minute (low stimulated saliva secretion) Score 2: < 0.9, ≥ 0.5 ml/minute (low stimulated saliva secretion) Score 3: < 0.5 ml/minute (very low, xerostomia).
'Buffer capacity' CRT® buffer was used. Immediately after the stimulated was collected as described previously, the dentist used a disposable pipette to place some of this stimulated saliva on the test strip.
After five minutes, the dentist compared the colour of the test strip with the standard colour chart. The scoring system for this parameter was performed as follows: 8 Score 0: High (normal or good buffering capacity) Score 1: Medium (less than good buffering capacity) Score 2: Low (low buffering capacity) 'Clinical judgement' Just before risk assessment and randomisation were performed for the first group of the patients, it was revealed that the calculated average of 'chance of avoiding new cavities' was higher than expected. Applying the increased risk score for the 'Clinical judgement' parameter, the average of 'chance of avoiding new cavities' is similar to that of an Arabian study for an adult population with a similar mean age and mean DMFS [5]. Possible reasons were as follows: 1. Almost all patients used both fluoridated water and fluoridated toothpaste, which converted to the most favourable score for the 'Fluoride programme' parameter.
3. The three-day food diary is self-reported and might lead to underscoring.
4. The reference data used for the 'Caries experience' parameter was from 15 years ago.
5. The eligibility criteria (medical-card-holder -proxy for low socioeconomic statuspatients who have 20 or more than 20 teeth) may not adequately capture the lower socioeconomic group.
For adjusting such systematic situations, the use of the 'Clinical judgement' parameter is recommended (Hänsel Petersson, G. personal communication, 16 December 2011). The current study complied with this recommendation. Additional file Table 3 summaries the distribution and mean (SD) of 'chance of avoiding new cavities' using both Score 1 (standard) and Score 2 (increased risk) for the 'Clinical judgement' parameter.