Cost-effectiveness of child caries management: a randomised controlled trial (FiCTION trial)

Dental caries has a large health and economic impact for the United Kingdom (UK) as it is the most common disease in children [1‐4]. Treating oral disease is expensive, costing NHS England £3.4 billion annually [5].

In the UK there is uncertainty surrounding the best strategy to manage caries in primary teeth, especially in primary care. There is debate about the clinical- and cost-effectiveness of conventional restorations (removing a carious lesion with a drill and placement of a restoration) compared to minimally-invasive biologically-orientated strategies (sealing-in a carious lesion with an adhesive restoration or preformed metal crown rather than removing it), or prevention-focused strategies [6‐9].

Cost-effectiveness analysis allows treatment comparisons in terms of both costs and effects [10]. Recent cost-effectiveness analyses of managing dental caries found the Hall Technique (HT), a method for managing carious lesions by sealing-in, to be cost-effective compared to conventional restorations [11] and compared to both conventional restorations and a Non-Restorative Cavity Control approach [12]. However, these studies followed outcomes on single teeth and have focused on one type of biological approach (i.e. HT).

A large trial, FiCTION (Filling Children’s Teeth: Indicated or Not?), was undertaken to measure the costs and effects, in terms of dental pain and/or infection, of three strategies to manage dental caries in the primary teeth of young children with dentine caries in the UK [13]. The strategies evaluated were Conventional restorations with best practice prevention (C + P), Biological management of carious lesions with best practice prevention (B + P), and best practice Prevention Alone (PA). The C + P strategy involved the complete mechanical removal of carious tooth tissue using local anesthesia and a drill followed by placement of a restoration alongside best practice preventive therapy and has been considered standard practice in the management of dental caries [14, 15]. The B + P strategy involved sealing-in carious lesions using a variety of techniques including adhesive restorative materials or preformed metal crowns placed using the HT along with preventive therapy; Schwendicke et al.’s (2018, 2019) analyses focused on the HT component of B + P [11, 12]. PA involved avoiding restorative intervention and using four components of preventive management; toothbrushing (with toothpaste of at least 1000ppmF concentration), dietary advice, fluoride varnish application, and fissure sealants to prevent further carious lesions.

The trial methodology and clinical outcomes are presented elsewhere [13, 16, 17]. In brief, this multi-center, three-arm, parallel group, patient-randomised controlled trial set in general dental practices in England, Scotland, and Wales was undertaken to evaluate three strategies to manage caries in the primary teeth of children aged 3 to 7 years with at least one primary molar tooth with caries lesions extending into dentine. The original planned follow-up was 3 years but due to an extension in the recruitment period this was revised to an average target follow-up of 35.5 months (a minimum of 23 months and a maximum of 36 months).

Reporting for this study follows the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) [18]. The trial was registered with the ISRCTN (reference number ISRCTN77044005) and East of Scotland Research Ethics Committee provided ethical approved (REC reference: 12/ES/0047).

A total of 1144 children were randomised and data on 1058 children were used in the economic analysis (n = 86 children did not have any clinical assessment of the primary outcome and were not included in the ITT analysis). The children included in the economic analysis were evenly distributed across the three arms in terms of numbers randomised and baseline characteristics; 352 randomised to B + P, 352 to C + P, and 354 to PA. On average, children were 6 years old [sd: 1.3] when recruited and there was an even split between females (51%) and males (49%). The CONSORT (Consolidated Standards of Reporting Trial) flow diagram is provided in Additional file 6 but additional baseline characteristics, and clinical findings are presented elsewhere [13]. The median follow-up was 33.8 months (IQR 23.8, 36.7).

The percentage of missing data for the economic analysis was low (< 5%). There were 7713 visits recorded across the three arms. On average, children had seven visits during their time in the trial, each lasting 21 min. All three strategies were similar in terms of average number of visits (mean visits [sd]: C + P 7.7 [4.2], B + P 7.4 [4.1], and PA 6.8 [3.7]) and duration of visits (mean minutes [sd]: C + P 21.8 [6.9], B + P 21.2 [7.2], and PA 20.1 [6.7]).

The number of visits at which preventive treatment was provided was similar across the three arms with slightly more prevention provided in the PA arm (C + P 79%, B + P 79%, and PA 85% of visits). The three strategies differed in the frequency of operative treatments provided, with less than 20% of all PA visits involving operative treatment compared to over 40% of B + P and C + P visits. The type of operative treatment provided also differed, as would be expected given the nature of the different strategies.

A total of 96 children (C + P n = 31, B + P n = 31, and PA n = 34 children) were referred on 107 occasions for additional consultations and/or further treatment (C + P n = 32, B + P n = 36, and PA n = 39 referrals) resulting in 52 general anesthetics (GA) being undertaken (C + P n = 15, B + P n = 12, PA n = 25 referrals with GA). Four children did not attend their referral appointment (n = 1 C + P, n = 3 PA).

Table 1 summarizes the average cost per child per visit for the three strategies.

On average, it cost £230 to manage dental caries in a young child with at least one primary tooth with a dentinal carious lesion over a period of up to 36 months. On average, C + P was the most costly and PA was the least costly strategy. Staff time, operative treatments, and patient referrals were the main cost drivers. As expected, C + P and B + P incurred more operative treatment costs compared to PA.

As PA was, on average, the least costly strategy we compared this to B + P, the next costly strategy, and lastly C + P was included in the comparison. In terms of effectiveness, there was no evidence of a difference in incidence, or in episodes, of dental pain and/or infection between the three strategies. Table 2 summaries the results of the incremental analysis.

B + P was, on average, more costly but more effective, in terms of both incidence of, and episodes of, dental pain and/or infection avoided, compared to PA. At a willingness to pay threshold of £330 we would consider B + P cost-effective to avoid an incidence and £130 to avoid an episode of dental pain and/or infection compared to PA. As C + P was, on average, more costly and less effective than B + P, in terms of both incidence of, and episodes of, dental pain and/or infection, it was dominated by B + P.

Figure 1 illustrates uncertainty surrounding the point estimates in Table 2. The figure presents the strategy with the highest probability of being considered cost-effective at each willingness to pay threshold to avoid an incidence of dental pain and/or infection.

Figure 1 illustrates that PA would have the highest probability (87%) of being considered cost-effective if a decision were to be based on cost alone. However, as the willingness to pay threshold increases, the probability of B + P being considered cost-effective increases, but it never exceeds 65%. C + P would not be considered cost-effective compared to PA and B + P in this analysis.

Figure 2 illustrates the strategy with the highest probability of being considered cost-effective at the different willingness to pay thresholds to avoid an episode of dental pain and/or infection.

In terms of episodes of dental pain and/or infection, the conclusions are similar except that B + P would be considered cost-effective at a lower willingness to pay threshold. The probability of B + P being considered cost-effective never exceeds 75%.

On average, it costs £230 to manage dental caries in primary teeth in a child with at least one tooth with a dentinal caries lesion over a period of up to 36 months. The main cost drivers were staff time, operative treatments, and patient referrals. On average, PA incurred a higher referral cost because that arm had more referrals and more referrals requiring GA, an important consideration in view of the morbidity associated with GA use [27‐29].

Although in terms of cost-effectiveness PA was, on average, the least costly treatment, it was also the least effective for both incidence of, and episodes of, dental pain and/or infection. There was an 87% probability that PA would be considered cost-effective as the least costly option but B + P and C + P would, on average, provide more oral health benefits, albeit at a higher cost. If society was willing to pay £330 to avoid one additional child experiencing dental pain and/or infection, B + P would have the highest probability (47%) of being considered cost-effective compared to PA (46%) and C + P (7%). Similarly, when society is willing to pay £130 or more to avoid an episode of dental pain and/or infection, B + P would have the highest probability (49%) of being considered cost-effective compared to PA (45%) and C + P (6%).

Vermaire et al. (2014) and Samnaliev et al. (2015) came to similar conclusions in their analyses, in that treatments aimed at caries prevention increased the cost of providing treatment and that the opportunity cost of these treatments is dependent on the payers’ willingness-to-pay [30, 31]. Our results differ from other studies in which the HT, which was a component of our B + P intervention, was reported to be more effective and less costly [11, 12] when compared to conventional and preventive based strategies. However, both of these studies by Schwendicke et al. (2018, 2019) were based on treating a single tooth, or two contralateral teeth per child whereas in our study the whole child/mouth (up to 20 primary teeth per child) could be treated, a situation more representative of real life treatment provision [11, 12]. Our study also had considerably more data available to inform our analysis (n = 1058 children, n = 2721 teeth; compared with n = 142 children and teeth in Schwendicke et al. 2018; and n = 91 children, n = 182 teeth in Schwendicke et al. 2019) [11, 12]. The costs estimated in the two Schwendicke et al. (2018, 2019) studies were based on current charges to the health system [11, 12]. In the present analysis we based our costs on a very detailed costing exercise, however, when we used current charges to the NHS in a sensitivity analysis we still reached the same conclusion (see Additional files 3, 4, and 5). Schwendicke et al. (2019) found a negligible difference in total treatment costs between HT and conventional treatment and this difference only became clinically and statistically significant when patient costs were considered [11]. Parental time and travel costs to attend appointments were not considered in our analysis. Inclusion of such costs are unlikely to change our conclusions as the average number of visits and length of visits were similar across the arms. Costs incurred by the parent and child due to toothache, such as time off work, childcare, and time off school, were considered in a sensitivity analysis but did not change our overall conclusions. In terms of oral health effects, direct comparisons could not be made with previous studies [11, 12] which considered the pain associated with dental caries together with endodontic treatment and extractions. The main differences in our study are, firstly, that B + P encompassed a number of minimally-invasive restorations of which the HT was only one and secondly, treatment was at the participant level (including all primary teeth) and not at the single tooth level.

This economic analysis had a number of strengths and limitations. The main strength was that the analysis was pre-planned and the data used were collected as part of the trial. There were few missing data and all available data were included in the analysis despite the varying follow-up. A limitation of the analysis was that SUR model may not have been an appropriate fit for the co-primary outcomes. However, a trade-off was made between fitting the most appropriate model and applying one that allows for the correlation of costs and outcomes, which the SUR approach does. Finally, capital costs were excluded from the analysis; this omission reduced the total cost of each arm equally hence the incremental costs, ICER, and our overall conclusions remain unchanged.

In practical terms, we do not know society’s willingness to pay threshold to avoid dental pain and/or infection in a primary tooth. A judgement is required as to what value the NHS places on avoiding dental pain and/or infection. Recent research conducted by Lord et al. (2015) estimated the willingness to pay to avoid dental caries with pain in a primary tooth [32]. They estimated this to be £153 (95% CI: £93 to £213 – inflated to 2017) [33]. If we adopted this as the willingness to pay threshold the PA arm would have a 68% probability of being considered cost-effective compared to B + P (29%) and C + P (3%) in terms of an incidence of dental pain and/or infection avoided. A willingness to pay threshold to avoid an episode of dental pain and/or infection also needs to be determined but based on the Lord et al. (2015) threshold, B + P would have the highest probability (53%) of being considered cost-effective compared to PA (40%) and C + P (7%). Further research is needed to identify the most appropriate threshold to assess our results.

To conclude, on average, PA is the least costly, despite having more referrals requiring GA, but the least effective strategy for managing dental caries in primary teeth. B + P has the potential to provide more oral health benefits to children with dentinal carious lesion in at least one primary molar tooth, however this comes at an additional cost. Over the willingness to pay values considered, the probability of B + P being considered cost-effective was approximately no higher than 65% to avoid an incidence of dental pain and/or infection and no higher than 75% to avoid an episode of dental pain and/or infection. It is unlikely that C + P would be considered cost-effective.