Introduction

Extraction of decayed teeth is the most common reason for UK children aged 5–9 years to receive a general anaesthetic.1 Over the last ten years these admissions have increased every year;2 in 2014–2015 over 46,000 under 16s were admitted for dental general anaesthetic (DGA).3 The burden of pain and disruption for children is large.4 The cost to the NHS is likely >£55 million (£1,179/procedure).5 These figures are conservative estimates, given problems of inconsistent and under-reporting of DGA in both hospital episode statistics (HES) and surveys to date.6,7 This problem is not restricted to the United Kingdom.8,9,10,11

Dental caries is largely preventable, and admissions could be reduced by prevention and early treatment. Sugar reduction, good oral hygiene, tooth brushing, and fluoride use reduce caries risk.12 The application of fluoride varnish and resin fissure sealant are key preventive actions.13 Current evidence suggests financial and organisational barriers discourage a preventative approach to oral health management by dentists.14,15

Oral health is recognised as a key health inequality16,17 particularly in childhood.1 Disadvantaged groups and individuals have worse oral health and less access to dental care.18,19,20,21 Those from more deprived groups experience both more decay,22 and more untreated decay.23 Despite free dental treatment for children in the UK, those from most deprived backgrounds visit the dentist less often and have higher rates of decay.24

Evidence regarding inequalities in tooth extractions under DGA is limited. The number of UK admissions is reported in HES,2 but these aggregate data offer limited opportunities for analysis by child characteristics or severity of decay. A recent study used community dental clinic data instead, and found rates of DGA were 2.5–5 times higher in the most deprived neighbourhoods compared to the least deprived neighbourhoods in Southampton, UK.25 Surveys of providers and attendees have established groups at higher risks of DGA, including those from more disadvantaged areas.27,28 Analyses of hospital records suggest children who are younger, have behavioural challenges or more severe disease are more likely to receive a DGA.26,27,29,30,31 In Western Australia, linked birth and hospital records have been used to show risks for DGA before the age of two years are higher among boys, indigenous children, those with intellectual disability, birth defects, and in areas without water fluoridation.32

To our knowledge, no studies have used hospital records to estimate rates of DGA in the population alongside estimates of disease severity, nor to examine area-based inequalities. Moreover, none have included area-based records of dental activity to examine possible associations between dental care activity and local rates of DGA. This paper aims to address these gaps by asking:

  1. 1

    What is the uptake and pattern of NHS dental activity for children in three local authorities in England?

  2. 2

    What is the burden of disease (incidence rate of DGAs and disease severity) for under-16s in the same area?

  3. 3

    Which individual or neighbourhood factors are associated with higher rates of DGA, and greater numbers of teeth removed?

Materials and methods

Three neighbouring local authorities in South West England with a combined child population >200,000 were included in this study. None of these areas receive fluoridated water.33 were included in this study. All routine referrals for DGA (ie, excluding children with complex medical needs and emergency procedures) from these local authorities are served by a single dental hospital.

We used two sets of routinely collected health data: National Health Service Business Service Authority (NHSBSA) and Bristol Dental Hospital (BDH) theatre records. NHSBSA data included a <18 population estimate. In addition we used Office of National Statistics estimates of <16 population per ward (2013 mid-year) for the population relevant to BDH records.34 Neighbourhood deprivation was estimated using the 2015 Index of Multiple Deprivation (IMD), averages for electoral wards as a continuous variable and as national quintiles.35,36

Data analyses were conducted using IBM SPSS Statistics v23 and Microsoft Excel 2013 for calculation of derived NHSBSA variables. All tests and P values were two-tailed. Choropleth maps were prepared using ggplot2 and ggmap in R-project.37

Reported dental activities: data set, preparation and statistical analysis

NHSBSA provided reported GDP NHS treatments in the year to March 2014. These data are recorded by the GDP at the close of each course of treatment (CoT). A single CoT may involve several visits (for example, examination plus treatment appointments). A patient may have more than 1 CoT per year if they visit a dentist more than once. NHSBSA had been previously cleaned and prepared for us by information analysts within the NHS, and were reported to us both by CoT and by unique patient numbers for all under 18s.

Data from NHSBSA was provided as anonymised, aggregated data sets, summarising activities by neighbourhood (electoral ward), patient characteristics and CoT for all <18-year-olds in the area. We estimated population treatment rates (number of unique patients receiving a treatment/100 <18s in area), treatment rate per 100 CoT (number of CoT including treatment/100 CoTs), and variation in CoT by patient characteristics (age, gender, ethnicity and home IMD).

Extractions under DGA: data set, preparation and statistical analysis

Bristol Dental Hospital (BDH) theatre records provided a patient level database for all children (aged <16 years, N = 1,509) receiving a DGA in the year to March 2014 matched to their hospital records.

Theatre records were accessed including the recording date of procedure, gender, trust number (a unique anonymous ID number), age in completed years, the number of teeth removed, and which teeth removed (total number of teeth removed was calculated where this was missing – where the total number of teeth removed was recorded as zero, these records were retained). Records were checked for eligibility and completeness and 31 ineligible or incomplete cases were excluded (Fig. 1). We supplied the BDH data intelligence team with 1,461 unique ID numbers, from which an anonymised data set comprising date of procedure, teeth extracted, calculated age at procedure date, home electoral ward, treatment details, and any admissions within the previous five years was retrieved. A total of 146 cases could not be matched to their BDH records. Data were checked for consistency and completeness. Date of birth data was complete, recorded home post code was missing or false for five cases and gender was discrepant in one case (treated as missing). Hospital treatment codes were used to check that these also showed dental extractions had taken place – theatre book records were assumed to be more accurate in this case – and a total of 30 records were excluded where neither dataset recorded an extraction (Fig. 1).

Figure 1
figure 1

Flowchart of record inclusion, Bristol Dental Hospital Records

Full size image

Patients from outside the three local authorities (Bristol, South Gloucestershire and North Somerset) which provided routine referrals, and those whose home local authority was missing were excluded before analysis (Fig. 1). We report here the local authority and ward level incidence rate of DGA; that is the number of <16-year-olds admitted to BDH for extractions under DGA per 1,000 population. These data were used to produce choropleth maps.

Ward level variation in DGA: data set, preparation and statistical analysis

We created a ward level data set from the NHSBSA data and ward counts of BDH cases.

Rate of DGA was analysed at neighbourhood (ward) level; the number of DGA cases originating in each ward was calculated and these totals, along with population estimates, were matched to the NHSBSA ward-level data set. We report the one-year incidence rate for DGA (the number of new DGA treatment cases as a proportion of average population during the period), acknowledging we do not have data on true disease incidence or prevalence during this period.

Pearson's correlation coefficients were used to assess associations between neighbourhood characteristics and ward-level incidence rate of DGA. Only those variables significantly correlated (P <0.05) with the dependent variables were used in a multivariable analysis. We undertook stepwise multiple regression adjusted for ward-level child population (model 1), child population and area deprivation (IMD) (model 2), child population, area deprivation and rates of treated decay (extractions and fillings per CoT), and child population, area deprivation, rates of treated decay and rates of preventive action (use of fluoride varnish, fissure sealant, and NHS treatment) (model 4) to ascertain the independent contribution of variables to the outcome. This enables us to comment on the extent to which the additional variables in each model improve our explanation of the data.

The number of teeth removed per child (a marker of disease severity) was analysed at an individual level as a continuous variable using ANOVA, and as a binary variable (<5 teeth removed, ≥5 teeth removed) in a binary logistic regression.

Patient involvement

This project was conducted as part of a Health Integration Team involving practitioners, local policy makers, third-sector providers, researchers and parents (www.bonee.org). Stakeholders were involved at every stage of the research, including planning, design and interpretation. Findings were disseminated to parents via BDH and local children's centres.

Ethics

The project was approved by the University of Bristol School for Policy Studies Research Ethics Committee. NHS permission for research was granted for us to access anonymous datasets from research and innovation at the University Hospitals Bristol NHS Foundation Trust (CH/2014/4,756) and the CaldEcott Guardian. The researchers did not have access to patient identifiable data.

Results

Reported dental activities

NHS dental activities were reported for 126,058 unique patients, representing 63 attendees per 100 <18 years in the population (Table 1). Among those who attend, mean attendance was 1.7 CoT in the year, and 17.3% received at least one application of fluoride varnish, 1.8% fissure sealant, 26.5% permanent fillings and 6.2% (or 7,789 children) extractions in the dentists' chair.

Table 1 Reported NHS dental activities by local authority reported, performance year 2013-14
Full size table

Dental activity per CoT in a one-year period by age group, gender, ethnicity and home IMD, are presented in Table 2. Children aged 3–18 years should be seen by a dentist and have fluoride varnish applied twice a year, those at higher risk of dental decay should be seen more frequently and have fissure sealants applied.13 Thus, we expect fluoride varnish a minimum of once per child. Fissure sealants should be used most commonly in the 5–9 age group. While 22.5% of CoTs among 5–9-year-olds include at least one permanent filling and 6.7% an extraction (most likely for caries in this age group), only 14.9% and 1.6% of CoTs include fluoride varnish and fissures sealant, respectively. Rates of permanent filling (25.2%), extraction (5.3%), fluoride varnish (20.8%) and fissure sealant (2.0%) per CoT were highest in the most deprived neighbourhoods. Rates of both treatment and preventive actions were higher in Black and Minority Ethnic (BME) than in non-BME patients, but ethnicity was unrecorded for a large number of patients so these data are not reliable.

Table 2 Treatment activities in one year period by child characteristic (not accounting for multiple treatments per patient)
Full size table

Extractions under DGA

Records of 1,509 procedures undertaken in the BDH general anaesthetic theatre were provided, with a total of 5,707 teeth removed.

Following matching and exclusions we include 1,120 recorded DGA in the year to March 2014 (Fig. 1). Most were young (median age 7 years) and most had >four teeth removed (Table 3). Of the 1,120 children receiving a DGA, 470 (42.0%) came from the most deprived neighbourhoods (IMD quintile 1), compared to 137 (12.2%) from the least deprived (IMD quintile 5).

Table 3 Characteristics of children presenting for dental extraction under general anaesthetic
Full size table

The one-year incidence rate of extractions under general anaesthetic was 6.6 for every 1,000 under 16 in the population, rising to 7.7 in the City of Bristol, and 12.4/1,000 among 5–9-year-olds (Table 3). Both the number (N = 629) and rate of DGA admission was highest in the City of Bristol, and the highest incidence rate was among 5–9 years old from Bristol (14.7/1,000 population).

Eighty-six (7.6%) of the children receiving a DGA in this year (for whom we could match records) had previously been admitted for this treatment. Seventy-seven children had received two DGAs, and nine children had received three or four. Among those repeat attenders, median age at first attendance was five years.

Ward level variation in rates of DGA

We compared the incidence rate of DGA between 106 electoral wards. Rates of DGA varied between zero and 20.9 per 1,000 child population (Fig. 2). Figure 3 shows IMD 2015 national decile for the same electoral wards, and Figure 4 the distribution of incidence rate by ward IMD decile where 1 is the most deprived. Incidence rate of DGA was highest in the most deprived wards and lowest in the least deprived wards, with 40% of attendances originating from wards in the lowest national quintile for IMD, but considerable variation between similar wards was observed.

Figure 2
figure 2

Ward level rates of DGA (number of admissions/1,000 child population)

Full size image
Figure 3
figure 3

Ward level national IMD decile (2015)

Full size image
Figure 4
figure 4

Ward level IMD decile (2015) and proportion of the child population experiencing DGA (mean and 95% confidence intervals)

Full size image

Neighbourhood deprivation correlated significantly with all measures of dental service provision tested (Table 4). More deprived areas and those with higher rates of all dental activities including treatment for decay and preventive actions had higher rates of DGA.

Table 4 Ward level variation in DGA incidence rate, NHS treatment uptake, and NHS treatments provided (N = 106 wards)
Full size table

Multiple linear regressions (Table 4) suggest that after adjusting for child population size, area deprivation and rates of preventive actions (specifically application of fluoride varnish) are associated with significantly higher rates of DGA (Table 4, Model 4, R2 = 0.8 SE 4.2). For every 1% increase in fluoride varnish application we see an increase in the incidence rate of 0.3–0.6 child DGAs per 1,000 children, and similarly for a one unit increase in IMD, an additional 0.3–0.5 children per 1,000 receiving DGA in a one year period. In contrast, including rates of treated decay (extractions and fillings) did not improve the model. When we include all our predictor variables, 82% of the variance in ward level incidence rate is explained.

Inequalities in disease severity were examined by analysing the number of extractions per patient (Table 5). Children aged 5–9 years had the largest average number of teeth removed per child (mean of 5.6 teeth, SD = 3.2) and those aged 10–15 years the smallest (mean of 3.7, SD = 3.3). We found no associations with any measure of area deprivation or service quality and the number of decayed teeth removed. The risk of having extensive decay (defined as five or more teeth removed) was significantly associated with child age only, where those in the age group 10–15 years were 70% less likely to have had five or more teeth extracted than the <4 reference group. The analysis of number of teeth removed per child as a continuous variable showed the same pattern of results.

Table 5 Average number of teeth removedper patient by patient characteristics (N = 1,099 children)
Full size table

Discussion

Summary of findings

We estimate that 40% of the child population did not visit a dentist in a one year period. Evidence suggests that preventive actions are underused by GDP in this region, with only 1/7 CoTs including fluoride varnish and 1/83 fissure sealant.

Around seven in every 1,000 <16-year-olds per year were admitted for DGA, of whom one in 13 were repeat admissions. Incidence rates were highest among 5–9-year-olds, in the most deprived neighbourhoods, and in areas with higher rates of fluoride varnish use. It is not possible to disentangle the causality and confounding in these observed relationships.

Most children had >4 teeth removed. Younger children had, on average, more teeth removed than older children. Number of teeth removed did not vary by gender, ethnicity or neighbourhood deprivation.

Results in context

Our data broadly agrees with previous studies reporting that 32.5% of children do not attend dental services,39 and that 8% of 3-year-olds and 25% of 5-year-olds have experienced dental decay in at least one tooth.40,41 Shaban et al. used the 2003 dental health survey data to

comment on inequalities in dental provision.24 Unlike us, they found higher rates of fissure sealant in less deprived groups (considering mothers education, family social class, and area deprivation). Their data is drawn from a smaller sample but is a more accurate estimate of total treatment incidence as it includes non-NHS treatment and treatment ever applied (while we only comment on application within the year by NHS provider) so is probably a more accurate estimate of total treatment incidence.24 Their data do not allow them to compare rates by child age.

HES data over the period 1997–2006 show around 30% of children admitted for DGA were from areas of high deprivation.2 In our sample, a larger proportion (40%) came from areas of high deprivation. This difference may simply reflect the different population samples; due to its geographic reach, Bristol Dental Hospital over-samples urban populations with higher rates of deprivation. We also report slightly different cases, we excluded children with complex healthcare needs who commonly receive DGA for reasons other than decay severity.42,43 Finally, our study reports cases seven years later than the previous publication, and it is possible that nationally observed reductions in rates of dental decay at age 541 have masked widening of inequalities if improvements have been concentrated in the most advantaged groups.

The young age of those receiving DGA in our sample confirms the young age at which caries is identified40,41 and dentists' preference for DGA for those they judge less likely to tolerate procedures under local anaesthetic including for the very young.44

We confirm the relationship observed by others between neighbourhood deprivation and rates of poor dental health in childhood.41 In particular, South Gloucestershire (the local authority with least deprivation in our sample) experienced low DGA rates echoing low rates of decay in national surveys.41

The observed association between higher rates of fluoride varnish use and DGA appears to contradict evidence of its role in reducing decay.45 In our population level study this probably reflects greater fluoride use in areas with higher rates of decay, and does not suggest fluoride use is ineffective. However, even assuming underreporting by GDPs, local rates of fluoride varnish application for young children are probably not sufficient to prevent the early decay most likely to result in DGA.

Given changes in practice, increasing rates of DGAs may not imply increasing disease rates.6 However, the high use of both DGA and permanent fillings among children studied here do represent large numbers of diseased teeth and the high burden of preventable disease in this population.

In studies of otherwise healthy children, prior DGA admissions are reported for 7.4% of children in a Leeds teaching hospital,46 and 5.6% of children using HES data.2 We report a rate similar to that found in the teaching hospital here, confirming the use of hospital records as a more accurate (and higher) record of repeat admissions and supporting the generalisability of the findings in Leeds and Bristol.

Strengths and limitations

The study strength lies in the accurate recording of attendance for DGA across one region, and in our ability to link individual records and area-based data. This allowed examination of area incidence rate and relationships to neighbourhood deprivation and dental activities. We were also able to comment on the number of teeth removed per child, which is rarely reported. This identifies the number of diseased teeth, but clinical and social features aside from severity of decay will determine whether teeth are removed or restored.

It is likely that our data still underreports DGAs. Children admitted for emergency surgery for acute dental pain, and the 11% of dental theatre records unmatched to hospital records are missing from the analysis. However, we believe these would account for a relatively modest number of missing records and have little effect on incidence rates.

Preventive actions by GDPs may also be underreported. GDPs may not accurately record all dental activities, particularly since fluoride varnish and fissure sealant do not generate additional payments.37 The study only reports activity within the NHS, so any private treatments received by children are not reported. Taken together these may substantially underreport GDP preventive activities, particularly in wealthier areas where more children will attend private practices.

In this observational study we are not able to attribute causality to the observed associations. Although we can comment on patterns of disease by neighbourhood, we cannot say with confidence what causes these differences. We also cannot comment on why these children received a DGA, so cannot differentiate whether this was due to young age, extent of decay, or dental anxiety.

Implications for policy, practice and future research

Our data show that the significant inequalities observed in child oral health are repeated in decay requiring extraction of teeth, although all neighbourhoods experienced DGA. These admissions are preventable, and follow disease established before the age of five years. The message for clinicians and policymakers is clear: prevention of severe dental caries should be a priority for child health in all areas. Children with dental pain present in dentists, community, primary and emergency health services and all could usefully play a role in follow up and prevention.47 Greater knowledge of oral health is needed among the entire child workforce.48

Despite Public Health England recommendations, the vast majority of NHS dental appointments for children did not include the preventive application of fluoride varnish. Few areas of England provide fluoridated water33 despite its known benefits for reduction of caries and tooth extractions33,45,49so this underuse of fluoride varnish is important. Use of fissure sealant was vanishingly rare (1.6%). Moving dental healthcare to a more preventive model is a significant national challenge.50,51,52,53,54 Policy makers must attend to the incentives (and disincentives) for preventive action, but individual clinicians can and should improve their practice. The UK Children's Oral Health Improvement plan and the increased role of local authorities in oral health promotion are positive moves, but research to support changes in professional practice is required.47,55

Research into avoiding readmissions for DGA is urgently needed. Fewer than half of those performing DGA have appropriate arrangements for ongoing preventive care.6 Studies have explored parents information needs post DGA,56 and there is a growing body of research into changing parental oral healthcare.57,58,59 But current clinical guidelines only cover post-surgery care, and make no mention of the need for post-surgery consultation with a general dental practitioner or other health professional.60

Conclusions

Too many children are receiving DGA in the area surveyed, and too few children receive preventive treatments from GDPs. Area-based inequalities in DGA incidence were apparent, but wealthy areas also experience significant disease burden. A preventive approach to oral health is needed, employing the full range of preventive options available including using dietary and fluoride interventions by GDPs and supported by all healthcare professionals to reduce dental decay levels and DGAs in children.