Introduction

Dental caries is a considerable global health problem affecting many children.1 According to the Global Burden of Disease study, caries in primary teeth affects 9% of children and is the tenth most prevalent health condition worldwide.2 Although Sheiham and James, among others, have argued that one of the reasons for the failure in combating this epidemic, is that insufficient attention has been hitherto paid to its primary cause – namely high sugar consumption,3 others have counter-argued that fluoride, particularly in toothpaste, makes dietary sugar control less necessary. However, there is now growing evidence that sugar is still a key player in dental caries incidence – even in areas where fluoride is widely available.4,5,6,7 Bearing in mind a growing national focus on efforts to limit dietary sugar consumption in order to address a growing obesity problem,8,9 it is now timely to consider how dentists address dietary issues when caring for children with dental caries.

National guidance from Public Health England in the form of 'Delivering better oral health,'10 forms the backbone of recommended evidence-based preventive dentistry in English general dental practice. This positions issuing dietary advice as central to the everyday clinical practice of general dental practitioners (GDPs); stating that: 'healthier eating advice should routinely be given to patients to promote good oral and general health'. The guidance is explicit, that the main dietary messages given should be 'to reduce both the amount and frequency of consuming foods and drinks that have added sugar.'10 It also mentions that there is a consensus that avoiding sugar-containing foods and drinks at bedtimes is beneficial to caries prevention, and draws attention to what a generally healthy diet looks like, in the form of an 'eatwell' plate. While the guidance does not tackle issues related to how GDPs should approach the delivery of advice, it does suggest that 'in some cases it can be helpful to use a diet diary', and an exemplar is given.10 This template used contains a three-day diary, with space to record time of consumption, alongside a space for a free text entry describing the item consumed.

In diet diaries, patients are typically asked to keep diet records for three consecutive days including at least one-weekend day. This detailed dietary assessment is meant to enable both a tailoring of dietary advice for individual patients, as well as prompting a detailed discussion between dentist and patient which opens channels of communication and thus makes the forming of a therapeutic alliance more likely.11 A further benefit envisaged, is that because diet diary keeping represents a real-time method of self-monitoring if used correctly,12 this can, in itself, effect a positive change in behaviour – for evidence shows that incorporating self-monitoring tools into behaviour change interventions increases their effectiveness significantly.13 While the individualisation of chair-side dietary advice has been widely advocated in dentistry for some time,14,15,16,17 it is somewhat surprising that such little evidence exists on the use of diet diaries in dental practice.

In healthcare more generally, there is concern that poor patient compliance can compromise the efficient use of diet-diaries.12,18 Completing diet diaries prospectively and in a timely manner over several days is essential for their successful use since these ensure the accuracy of the record and its representativeness of habitual intake.18 Therefore, non-compliance with diet diaries usage does not just concern the problem of a patient failing to return with a completed diary, but also where patients fail to keep a full, contemporaneous account of their diet and return either incomplete diaries or those which have been backfilled.19 Missing information may undermine the validity of diet diaries and breach the rationale for their use, which is to enable a tailoring of dietary advice to the patient's dietary behaviours. Of course low completion rates of diet diaries may also be due to a failure of clinicians to issue them to patients in the first place.

A recent study of the use of diet diaries in general dental practice, suggests low compliance might be an issue, although the matter was overshadowed by practitioners' perceptions that the use of diet diaries in NHS dental practice is insufficiently supported by the current system of dental remuneration.20 Therefore, in order to look more closely at issues to do with patients' compliance with the use of diet diaries in dentistry, and whether useful information is yielded by their use, we set out to study the use of diet diaries in a dental setting where clinician remuneration is not an issue – a teaching hospital situation, where care protocols currently advocate their use. The primary aim of this study was therefore to investigate the return rate of diet diaries issued to child patients aged 5–11 years old in a teaching hospital setting, and whether this was associated with certain demographic or oral health-related factors. A secondary aim was to look at what type of information was yielded by diet diaries, which had been completed and returned, in order to investigate the extent to which this tool is capable of capturing the variety of dietary behaviours relevant to developing dental caries.

Methods

Ethics (reference 14/LO/1204) and NHS research governance approvals were obtained before commencing the study.

Setting and sampling

The study was carried out in the paediatric dentistry department of Liverpool University Dental Hospital (LUDH) – a teaching hospital that provides secondary dental care for children referred from NHS dental practices in the region. All children/parents with dental caries attending this centre are routinely asked to keep diet diaries as part of their caries risk assessment, at the start of their care. Clinic data shows that on average, 10% of children fail to keep this first appointment. Twenty percent of those attending the first appointment also fail to attend a subsequent appointment when the diet diaries would be reviewed. In Liverpool clinicians routinely complete prevention pro formas (information of social and dental history and oral health behaviours) for all these patients as part of the caries risk assessment process, and so completion of a prevention pro forma in the clinical record, was taken as an indicator that a diet diary had been issued. We retrospectively evaluated a random sample of records of children aged 5–11 years attending prevention clinics in LUDH between January 2010 and December 2013. According to the teaching hospital database, 519 eligible children attended during this period. Based on previous audit information suggesting that 30% of these patients returned the diet-diaries, a sample size of 200 records was identified as sufficient to allow an estimate of the proportion of returned diet diaries to within 5%, given predicted 95% confidence intervals. Using record tracking codes, a random sample of 200 clinical records was selected. When retrieved, only 174 of these clinical records contained completed prevention pro formas and so only these records were included in the study

Data extraction from clinical records

The following information was retrieved from the prevention pro formas in case notes: social history (child's age, gender and post code of residence, number of siblings and parents in the household; and whether the grandparents lived with them in the same home); DMFT–dmft (number of decayed, missing and filled permanent and primary teeth); oral hygiene practices (tooth brushing was considered regular if reported to be twice or more per day); and dental attendance (if patients reported a dental visit of any kind within the last six months, this was considered regular; if not, dental visiting was considered irregular). Postcodes were used to identify the corresponding Index of Multiple Deprivation quintiles (IMD),20 ranging from Quintile 1 (the most deprived) through to Quintile 5 (the least deprived). Any completed diet-diaries included in clinical records were photocopied after anonymising the patient's identity.

Content analysis of diet-diaries

In order to address objective 2, a content analysis was applied to dietary information contained for all completed diet-diaries. This form of analysis requires the use of a coding framework, comprising categories with conceptual definitions, in order to inform the identification and classification of data.21 We undertook this analysis deductively using a coding framework which had been previously developed from the dental literature17,18,23,24,25 and an earlier study involving GDPs responses to a diet diary vignette.26

Content from photocopies of diet-diaries was transcribed verbatim before being transferred into NVIVO software-version 10, to facilitate coding and analysis. The coding framework used contained 11 aspects of dietary assessment previously identified as potentially relevant to the giving of dietary advice by GDPs (Table 1). To help ascertain whether a food/drink item should be coded as harmful, or containing hidden sugars, information in the diary was at times supplemented by referring to UK food tables and nutritional labels on market websites. Each of the 11 aspects of dietary assessment was coded as being missing from the diary when there was either no description (defined as complete lack of relevant information) or insufficient description (defined as limited information rendering the data misleading or judgement impossible) contained in the diary for that category.

Table 1 Coding framework used in the content analysis of diet diaries and frequency of information extracted from diet diaries
Full size table

Data were coded by one investigator (AA) and verified by another independent assessor in the first 15 (25.0%) diet-diaries. Both assessors were qualified dentists. Cohen's κ was run to determine the level of agreement between the two coders. There was strong agreement in most of the codes, κ = (0.72 to 1.00), P ≤0.05. Frequencies of these codes were obtained by simply counting the observations of each code.

Statistical analysis

Data management and statistical analysis were performed using statistical software SPSS Version 22.0. (Armonk, NY: IBM Corp.). Descriptive statistics were used to profile the study sample and to describe the return rate of diet-diaries, with comparisons across socio-demographic characteristics, oral health related practices and dental caries experience of the study sample using the Chi-squared test and the Mann-Whitney U test. Binary logistic regression models were used to identify potential predictors of returning diet-diaries. A significance level of 20% in univariate analyses was used as a cut-off point to include variables in the multivariate model.

While a cut-off for the acceptable proportion of missing data in a data set that allow valid statistical inferences to be made has never been established, it has been suggested that if more than 10% of data is missing, this is likely to cause biased analysis.27 Nearly 25% of values related to family size, number of siblings, household parents and grandparents were found to be missing from prevention pro formas. So in order to explore the potential impact of missing data, an additional sensitivity analysis using multiple imputations (MI) was performed. Five complete imputed data sets were created using the fully conditional specification method, assuming that values were missing at random.28 The results from the analyses of the five imputed data sets were combined to give pooled estimates for the effects of the predictors.

Results

Of 200 records sampled, 174 had prevention pro formas, and were included in the analysis. Diet diaries were found in 60 out of 174 records, giving a return rate of 34.5% (95% confidence interval 27.4% to 41.6%). The socio-demographic and oral health-related characteristics of the sample, as well as a comparison of diet-diaries return rate by these characteristics are shown in Table 2. The final study sample comprised records of children with a mean age 7.2 (±1.7), and mean DMFT–dmft of 6.1 (±2.8). While the majority of sample was from areas of the most deprived IMD quintile (121, 69.5%). The majority of the sample also reported regularly visiting the dentist (129, 74.1%) and regularly brushing their teeth (132, 75.9%). The sample also had an even gender balance; with a roughly equal balance also, between single parent/dual parent households.

Table 2 Study sample description and bivariate comparisons of diet diaries return rate (N = 174)
Full size table

Bivariate comparisons of the return rate of diet-diaries by sample characteristics showed a higher return rate among regular brushers (P = 0.016) and small families (P = 0.035), Table 2. These differences disappeared in the adjusted multiple regression models. Yet, when the multiple imputations were applied, adjusted models showed that regular brushers were more likely to return diet-diaries (Table 3).

Table 3 Multiple regression analysis of factors associated with diet diaries return rate
Full size table

Content analysis of the 60 completed diet-diaries is presented in Table 1. It was possible to identify consumption of 'harmful items' in all diaries, and also the frequency and between-meals consumption of sugars, as well as general aspects of the diet, from more than 90% of diaries. However, information on the sequence of food/drink consumption could not be elicited at all, and information on sugar amount and prolonged contact with teeth could only be identified from just over half of diaries. The least frequently extracted information was the context of the intake (17, 28.0%) and whether the consumption was near bedtime (18, 30.0%) (Table 1).

Discussion

Firstly we must recognise the study limitations such as missing data which are inevitable in this type of study.29 Some case notes were excluded from the final analysis because they did not contain prevention pro formas, and this effectively reduced our sample size a little. Nevertheless, given the confidence intervals involved in the study, we were able to establish the diet diary return rate at a precision rate of ±6%; although we have also made the assumption that completion of the prevention pro forma meant that a diet diary had been issued, and we have no independent verification of this, and so this should also be borne in mind when interpreting findings.

Our study shows that a relatively low proportion of diet-diaries are returned (34.5%, 60), even in a dental setting where NHS remuneration is not an issue. This study also found that regular brushers, and children from small families, were more likely to return diet diaries. Since regular tooth brushing is a sign of the patients' and parents' motivation and positive attitudes towards oral health, and a smaller family size probably indicates that these families are those with more time to complete a diary task, collectively our findings suggest that many of those children/parents who would benefit most from detailed dietary advice for caries prevention, lack the necessary motivation and time to comply with their usage. The low return rate of diet-diaries observed in this study, may also have something to do with the fact that the majority of records included in this study were related to patients from the poorest end of the socioeconomic status (SES) spectrum. Given that patients' poor compliance in health care settings is not uncommon, for example, non-compliance rates of 19–28% for attendance at medical appointments30 and 30-50% for treatment regimens31 have been reported, a compliance rate of 35% for diet diaries usage among a predominantly low SES is not unexpected bearing in mind the motivation required to complete the diary task. In addition, diet diaries are found to be the least preferred method of dietary assessment by people from low SES backgrounds when compared with other diet assessment methods; owing to low literacy, numeracy and language skills.32 Given that our sample were also those who evidently experienced high rates of dental caries, it is a concern that those most in need of help with caries prevention, appear not to find the current approach to dietary assessment methods appealing. With caries now focused among low socio-economic groups in many countries,33 our study therefore calls into question whether the routine use of diet diaries in dentistry is appropriate.

A second objective of the study was to examine the quality of information yielded by completed diet-diaries, which could be seen as relevant to informing patients' dietary advice aimed at preventing dental caries. Our content analysis of completed diaries showed an important and clinically relevant finding: that diet diaries did not consistently capture the full range of complexities relevant to giving dietary advice to dental patients. Information on sugar amount, consumption context, sequence of intake within meals, prolonged contact with teeth and sugars consumed near bed-time – all of which are recognized as being detrimental to developing dental caries,15,23,24 was often partially or completely missing from returned diet-diaries. Most striking on this list is that amount of sugar consumed could not be extracted from many diaries, even though national dental guidance10 articulates this consideration as one of the main dietary messages which should be covered when giving advice. This is of particular concern given very recent evidence which indicates that dietary sugars amount may be a more important factor related to caries development, than frequency of sugar intake.6 Although it is beyond the remit of the study to explain the reasons why information on sugar 'amount' is not given by patients completing diet diaries, we can surmise that there are two possibilities – either patients see this as unimportant, or too bothersome to report,34,35 or that dentists, when issuing the diet diary, place more emphasis on recording behaviours that they personally perceive as particularly important (and effectively prime their patients). Since we know that dentists see reducing frequency of sugar consumption as more practical than reducing the amount,23 the second explanation here is a real possibility.

'Delivering Better Oral Health' guidance10 also identifies that identifying 'hidden sugars' in the diet is a key objective. It is important, therefore, to recognise that in our study, extraction of data from diet diaries allowed the seeking of additional information from other sources such as food tables where hidden sugars/harmful dietary items required clarification. A GDP undertaking a chair-side interpretation of diet diaries would probably not be realistically able to also drawn on these additional resources. On the other hand, one could probably argue that what would happen in clinical practice in the event of such uncertainties, is that the GDP would probe for this missing information during a chair-side discussion of the diary with the patient. However, relying on further prompting then effectively compromises the prospective nature, temporal proximity of recording and accuracy of recording dietary intake, which are all, considered to be the major strengths of diet diary usage.18 Moreover, given that distinguishing a dietary record that reflects actual consumption from dishonest or incomplete ones is seen as nearly impossible,36 because diet diaries are often subject to recall bias18 and patients are known to tend to simplify or alter recordings to avoid negative feedback,37 we are left asking – are diet diaries any use?

Thus we may need to look for more modern tools to support dietary assessment in dentistry. Certainly digital technology offers new, more modern alternatives to the traditional paper diet diary record. Such alternatives may enhance patients' compliance and engagement in dietary assessment and self-monitoring activities by reducing the participant burden.38 A recent systematic review of health interventions, including dietary interventions administered using mobile apps, indicates that mobile apps supporting the self-monitoring of health related data have a high level of feasibility and acceptability.39 In line with the explosion of innovation in this area, Public Health England has recently launched the 'sugar smart app' which helps users to recognise total sugar in different dietary products, as part of the Change4Life advertising campaign.40 However, while this app may be helpful in identifying the presence of sugar in diet, it still fails to account for the complex nature of the association between other aspects of sugar consumption behaviours and dental caries.

Finally, bearing in mind the social gradient of dental caries distribution, with higher prevalence in children from socially deprived groups,33 such technology-based interventions need to be evaluated for their affordability and accessibility for these high risk groups and also for different age groups. Until this happens, working on more upstream approaches to tackle sugar consumption related to dental caries prevalence in these groups could be a better option.

Conclusion

The return rate of diet diaries by children and their families in a dental hospital setting is low, and appears to be associated with patients' demographic and oral health maintenance habits. Returned diet diaries showed a varied range of frequently missing, important dietary information. This included the amount of sugar consumed which is one of the main aspects of the diet recommended to be covered by dentists in national guidance, and thus effectively compromises the validity of diet diaries as a diet assessment tool for everyday clinical practice. This then raises questions as to whether this tool is the most appropriate means to support caries prevention for groups most in need of advice.