The study had two main aims:

1. to compare the clinical effectiveness of the Hall Technique with conventional methods of managing carious primary molars; and

2. to compare the acceptability of the Hall Technique to children, their carers, and dentists with conventional restorative methods for carious primary molars.

The null hypotheses tested in this paper are that after two years, for restorations placed by GDPs when managing caries in primary molars in this increased caries risk population:

• there is no difference in the incidence of signs and symptoms of pulpal disease between teeth restored with Control restorations and those with PMCs fitted using the Hall Technique;

• there is no difference in longevity of restorations between Control restorations and PMCs fitted using the Hall Technique; and

• children, their carers and GDPs have no preference between the Hall Technique and the conventional restorative care provided by their GDPs.

The study was a randomized controlled clinical trial, with a split mouth design set in general dental practice in Tayside, Scotland (2000 regional dmft 2.47, d₃ 1.71, mt 0.54, ft 0.22 33). Approval from the Tayside Committee on Medical Research Ethics (Ref 108/00) was obtained. Written and verbal consent from parents and, where appropriate, the children, was obtained, following a minimum 24 hours "cooling off" period between their being given information, and consent being obtained.

The GDPs recruited children who had caries affecting pairs of primary molar teeth, which were matched for tooth type, arch and extent of caries. Computer generated randomisation for sequence and side were held centrally, and accessed by telephone to a distant co-ordinator prior to treatment. The tooth on one side was restored using the Hall Technique and the contralateral tooth with the restorative technique the GDP would normally use. Data relating to the two treatments were recorded at the treatment appointment and follow up data at yearly intervals.

A systematic review of 10 studies 15, comparing the performance of PMCs with amalgam restorations gave a mean failure ratio of 0.32 in favour of the PMCs. Using this data, a two tailed test (α = 0.05), and a power of 80%, it was calculated that a clinical trial would require 58 participants to demonstrate a clinically significant difference in treatment outcomes of 50%. Allowing for loss of participants over a two year period, the target sample size was 120, with a maximum of 200 patients. Sequential analysis was to be undertaken when 90 patients had returned for one year follow-up. If this indicated a significantly increased failure rate for the intervention, recruitment would cease before the maximum of 200 patients.

All 143 GDPs in Tayside, Scotland were invited by letter to participate in the study. From the 41 replies expressing an interest (29% positive response rate), 17 GDPs were selected to obtain a spread of dental practices throughout the region. The GDPs attended a training session at Dundee Dental Hospital, with discussion on the study protocol, training in standardised bitewing radiography techniques using size 0 (22 × 35 mm) films (ULTRASPEED films, Eastman, Kodak, Hertfordshire, UK), Rinn holders (Dentsply, XCP film holding system; Rinn Corp, Elgin, Illinois, USA), and instruction in the Hall Technique.

Each dentist was asked to recruit 10 patients to the study who had caries affecting pairs of primary molar teeth, which were matched for tooth type, dental arch and extent of caries. Inclusion criteria required children to be 4–9 years of age, with no significant health problem, and presenting for routine dental care to their GDP. Patients had bitewing radiographs taken before entry to the study and then at annual recall visits. Details of patient recruitment and follow-up numbers are listed in the Consort statement (Figure 2).

For inclusion within the trial, teeth had to be:

• pairs of unrestored carious primary molars;

• matched for tooth type, dental arch, and extent of caries (radiographically ≤ or >1/2 way through dentine); and

• symptomless, with no clinical or radiographic signs of pulpal pathology on bitewing radiography as assessed by the GDP.

Where more than one pair of matched carious lesions were present in a child's mouth, the dentist chose which pair should be part of the study. Any carious teeth outwith the study were managed as per the dentists' normal treatment regime. Figure 3 shows the radiographs of a pair of teeth entered into the trial.

The dentists used their discretion to decide whether the two restorations were carried out at the same, or separate appointments, and recorded their decision. The protocol for the provision of the control restoration (Control restoration) followed currently accepted practice, requiring complete removal of caries from the periphery of the cavity, and as far as possible from the base of the cavity without causing pulpal exposure. Local anaesthesia was to be administered if the GDP would usually use it in that clinical situation, and the cavity restored with the material the GDP would normally choose.

The protocol for provision of the Hall Technique PMC (Hall PMC) was as follows:

• obvious food or debris were removed from the cavity but no caries removed;

• the child was positioned upright in the dental chair to reduce the chance of accidental swallowing or aspiration of a loose PMC. Additional airway protection could be gained by use of a gauze swab behind the tooth or by securing the PMC with a strip of elastoplast;

• the correct size of PMC for the tooth was selected i.e. covering all cusps and giving the feeling of "spring-back" when placed up to, but not through, the contact points. If the contacts were very tight, orthodontic separator elastics could be used and the Hall PMC fitted at a subsequent appointment;

• the tooth was rinsed and dried, and the PMC dried;

• the PMC was filled with glass ionomer luting cement. If the cavity was large, some glass ionomer could be placed in the base of the cavity before cementing the crown;

• the PMC was placed evenly over the tooth and the child instructed to bite down firmly until the crown was pushed down over the tooth;

• if the child was unable or unwilling to bite down on the PMC, finger pressure was used to seat the crown;

• extruded cement was removed, and the child asked to keep biting on the Hall PMC until the cement has set; and

• once cement had set, excess cement is removed, floss was used to clear the aproximal contacts, and post-fitting instructions were given.

Immediately following each treatment, the GDPs were asked to record the following information on a form:

- any difficulties encountered with either treatment;

- whether caries removal for the control tooth was complete or incomplete (and if incomplete, why);

- the material used for cavity restoration in the control tooth;

- the GDPs subjective assessment of the discomfort level experienced by the child during each procedure, using a five point scale (descriptors listed in Figure 4);

- the distance, measured in millimetres at the incisors, that the Hall crown caused the occlusal vertical dimension (OVD) to be increased by;

- whether the restorations were carried out at the same or separate treatment appointments; and

- the time taken for explanation, and the duration of each treatment.

Following recruitment and consent, if either restoration could not be placed, then the patient continued to be included in the trial and monitored as per protocol (following the principle of "intention-to-treat"). On completion of both treatments (see Figure 5), the GDP asked and recorded which treatment option the child, their parent/carer (if present), and they had preferred.

Patients were kept under normal review intervals by their dentists, with clinical and radiographic data (from bitewing radiographs) being recorded at yearly intervals. Details of emergency visits were also recorded. At each annual recall, a pre-printed form was completed, recording:

• a full dental chart;

• whether there was occlusal contact on both sides of the arch;

• whether the child had experienced any temporomandibular joint (TMJ) related pain or difficulty with eating;

• any dental pain the child had experienced and whether the child had required emergency dental treatment. Any emergency treatment required by either of the study teeth outwith routine recall was also recorded; and

• the success or failure of the restorations, assessed clinically and radiographically.

GDPs monitored the success or failure of the restorations using their usual clinical criteria as to whether a restoration required a further intervention, such as repair, replacement or if the tooth required pulp therapy or extraction. If there was a dental intervention, then they recorded this on their pre-printed annual recall form. Two researchers (NI and DE), reviewed these forms, and the GDPs radiographs, and using the criteria in Table 1 ascribed the following outcomes:

• Successful;

• 'Minor' failure – restoration failure, or reversible pulpitis, which could be managed by repair or replacement of the restoration (see Figure 6); and

• 'Major' failure – signs or symptoms of irreversible pulpal damage, such as dental abscess, or tooth broken down and unrestorable (see Figure 7 for an example).

These outcome criteria were chosen on the basis that they provided pragmatic, relevant information for clinicians and patients.

Data were entered into an Access database (Microsoft XP Professional, Microsoft Corporation) and analysed using Minitab 14 (Minitab^® Statistical Software, Minitab Inc.).

Radiographs were analysed by one author (DE) using a standard illuminated radiograph viewer, screened off, in a darkened room and using two times magnification. Data recorded from the initial radiographs included:

• quality of radiographs for; exposure/developing, visibility of study tooth crown, visibility of study tooth furcation area;

• classification of the most significant carious lesion affecting the tooth as either occlusal (requiring a Class I restoration) or aproximal (requiring a Class II restoration);

• caries less than/equal to half way through dentine or greater than half way through dentine; and

• presence or absence of signs of pathology at furcation region (this criterion was only assessed if the inter-radicular area was recorded on the film, and close proximity of the successional permanent tooth did not confound the diagnosis).

For radiographs taken at recall visits, additional criteria included:

• whether the fit of the PMC was satisfactory (crown margin through both mesial and distal contact points);

• presence or absence of a satisfactory Control restoration; and

• any evidence of caries progression or of new carious lesions.

Investigation of inter- and intra-examiner reproducibility and repeatability were carried out using a computer generated randomisation table to select 10% of individual radiographs, which were then reassessed by two of the authors (DE and NI). Kappa analysis values ranged from 0.60 to 0.84 ('good' to 'very good').

Data gathered from all proformas were entered into an Access database (Microsoft XP Professional, Microsoft Corporation) and analysed using Minitab 14 (Minitab^® Statistical Software, Minitab Inc.). Chi square test was used to analyse data on:

• relationship between the extent and position of the initial carious lesion and the occurrences of the outcomes of pain episodes, 'Major' failure and 'Minor' failure;

• failure of Class I GIC restorations compared with other restorations and failure of Class II GIC restorations compared with other restorations.

• patient, parent and child preferences for the control or Hall restoration; and

• patient preferences and the influence of whether the control or Hall restoration was carried out first or second.

McNemar's test was used to analyse paired data on:

• pain episodes, 'Major' failures and 'Minor' failures in the control group and the intervention group.

17 dentists recruited 132 patients over a period of two years six months (July 2001 to January 2004). Each dentist recruited between 0 and 21 patients, with the distribution shown in Figure 8.

69 males and 63 females were recruited from three to 10 years of age (mean 6.8 years; SD 1.58) with the distribution shown in Figure 9. Two children were three years of age and six children were 10 years of age. In the inclusion criteria, the lower age of four years was originally chosen because it was felt that children any younger would not tolerate radiographic examination. This decision was not based on clinical rationale relating to the treatments. It was, therefore, decided to include these patients. With the patients over nine years of age, the recruiting GDPs anticipated that the teeth were still likely to be present after two years. As the study was a split mouth design, inclusion of these teeth would not bias either for or against the intervention or control and they were, therefore, also included in the analyses.

132 matched pairs of teeth were entered into the study with the distribution being evenly spread by tooth type (first or second primary molar) and arch (maxillary or mandibular), and is shown in Table 2. Seventy three study teeth pairs (55%) were first primary molars and 59 (45%) were second primary molars.

There was no radiographic information for 56 teeth out of the 264 entered into the trial (31 had no radiographs available; radiograph quality too poor for assessment for 10 teeth; lesion too minimal for site to be confirmed for 15 teeth). However, for 21 of these 56 teeth with no radiographic information on the initial lesion, the GDPs dental chart and the location and extent of subsequent restorations viewed radiographically were used to add information regarding initial lesion characteristics. Therefore, data regarding site and extent of the initial lesion were available for 229 of the 264 teeth entered into the trial (86%). This comprised 115 Hall teeth and 114 control teeth, with similar distributions for location and extent of caries lesions (Figure 10). A mean of 58% of lesions extended ≤ 1/2 way into dentine and 42% >1/2 way into dentine, with 32% of the lesions on the occlusal surface and 68% on an aproximal surface.

Control restorations were placed in 128 teeth out of the 132 allocated to the control group (97%). For the four teeth where no restoration was placed, three children were unable/unwilling to receive any treatment and for one patient, no reason was given. Table 3 details whether complete or incomplete caries removal was carried out and the type of restoration provided. 103 of the 132 teeth in the sample (78%) were recorded as having complete caries removal carried out. The spread of restorative materials chosen by the dentists comprised: glass ionomer cement (69%); amalgam (8%); compomer (5%); composite (11%); PMC (1%); and fissure sealant (2%). Four teeth (3%) had no restoration provided. The one PMC that was placed was fitted in error as there was confusion around the randomisation process. The dentist prepared the tooth before fitting the PMC, although no caries removal took place. The reasons for incomplete caries removal in the Seventy three study (22%) scheduled for Control restorations are shown in Table 4. An air rotor handpiece was used for cavity preparation in 114 out of the 128 teeth (89%) where some treatment was carried out.

PMCs were placed using the Hall Technique on 128 out of 132 teeth (97%) allocated to the intervention arm, with the fit being assessed by the GDPs as satisfactory for 124 of the 128 teeth (97%). All four teeth where no Hall PMC could be fitted were maxillary first primary molars and the reasons for a PMC not being placed are shown in Table 5. The four cases where a PMC was not fitted and the four cases where no restoration was placed involved a total of six patients. In two patients, neither treatment could be provided.

Of the 128 teeth where Hall PMCs were fitted, 119 (93%) had at least one set of bitewing radiographs taken at follow-up appointments, and from which the radiographic appearance of the Hall PMC fit could be judged. Radiographs were not available for nine teeth where Hall PMCs had been successfully fitted as two patients did not attend any follow-up appointments, one tooth exfoliated before the next follow-up appointment where radiographs would have been taken, and six teeth had no follow-up radiographs for unknown reasons. The fit of the Hall PMC was judged from the radiographs to be unsatisfactory if one or both proximal margins of the Hall PMC were not cervical to the contact points (see Figure 11 and 12), and this was the case for 18 of the 119 (15%) teeth. Seventeen out of 18 of the teeth were first primary molars. All four PMCs assessed as being an unsatisfactory fit by the GDPs were included within the 18 PMCs assessed by the authors (DE and NI) as being an unsatisfactory fit radiographically.

Orthodontic separators were used for 17 cases out of the 128 intervention teeth (13%) where PMCs were successfully fitted, but their use was not evenly distributed amongst GDPs in the trial, with only seven out of the 16 GDPs who fitted Hall PMCs using them. Four of the dentists who used separators did so for only one case; one dentist used them for six out of their eight cases, with the final dentist using them for seven out of the eight cases they enrolled in the trial. Table 6 shows the relationship between use of separators and whether the Hall PMC was assessed as fitting satisfactorily or not. A Chi square test showed no evidence of association (p = 0.810) between use of separators and satisfactory Hall PMC fit.

For 89% of Hall PMCs, procedures were rated by the GDPs as causing "no apparent discomfort" to "mild, not significant", while for Control restorations, the figure was 78%, and the difference was statistically significant (Chi square test, p = 0.012). The distribution of this data is shown in Figure 4.

Following completion of both procedures, the dentists recorded which procedure (Hall PMC or Control restoration) they, the child and their carers had preferred. Over nine out of 10 children and dentists, and three-quarters of carers, expressed a treatment preference (Figure 13). For 77% of the children, 83% of carers and 81% of dentists, the preference was for the Hall PMC, and this was statistically significant (one sample Chi square goodness of fit test; P < 0.0001) for all three groups. Therefore, the null hypothesis that children, carers and dentists would not have a preference was rejected.

The use of separators did not influence either the level of discomfort experienced by the child when having a Hall PMC fitted, as assessed by their dentist, or the preference for either technique expressed by the child, their parent or their dentist. When the Control restoration and Hall PMC were provided during the same appointment, the children's preferences were not dependent on which restoration was carried out first (Chi square analysis; P = 0.203).

As the Hall Technique does not involve any occlusal reduction of teeth, it is inevitable that the placing of a Hall PMC will increase the occlusal vertical dimension (OVD). GDPs measured this increase at the incisors immediately after placement of the Hall PMC. The mean reported value for all teeth was 2.4 mm (SD 0.13, range 0–4 mm). For first primary molars this was 2.3 mm and for second primary molars the mean value was a little higher at 2.5 mm. Even occlusal contact was recorded on both sides of the arch for all 129 children at the one year recall appointment. An example of a patient treated with Hall PMCs demonstrating re-established even occlusal contact is shown in Figure 14.

There was no history of TMJ pain, and no difficulty with eating reported by either the child or their parent for all 129 children assessed at one year recall.

For 84 patients (64%), the treatments took place at the same appointments and for 48 (36%), at separate appointments. The time taken to explain and complete each procedure was reported for 129 Hall PMC procedures and 128 Control restoration procedures. For the Control restorations, a mean time of 11.3 minutes (range 4 to 32 minutes; SD 5.5) was reported, with a mean time of 12.2 minutes (range 2 to 40 minutes; SD 8.3) for the Hall PMCs. Table 7 shows a breakdown of the times for each procedure.

The minimum period of 23 months (rather than 24 months) was chosen to take account of minor variation in dental practices arranging yearly recalls. Data were analysed for all patients with follow-up data at a minimum of 23 months (mean 28 months), for the period 0 – 36 months following enrolment into the trial. Patients with data only at less than 23 months were not included in the follow-up data analyses. Of the 132 patients enrolled in the study, there were 124 patients (94% follow-up rate) who met these criteria. Table 1 details the outcome criteria for the clinical and radiographic assessment of the restorations and teeth.

For the period of 0 to 36 months, there were a total of 422 follow-up radiographs (211 Control and 211 Hall) available for the 124 patients meeting the inclusion criteria, as there was more than one set of radiographs for some patients. There was radiographic information for 117 patients out of 124 (94%) where this might have influenced the assessment of success or failure (one patient's teeth exfoliated before the one year recall and six had no follow-up radiographs for unknown reasons). The results for radiograph quality assessment are shown in Table 8. The presence or absence of new caries or caries progression could be determined (codes 0 and 1 for exposure/developing; codes 0 and 1 for visibility of study tooth crown) for 114 Control restoration allocated teeth (88%) and 117 Hall PMC teeth (91%). Information on the presence or absence of inter-radicular radiolucencies (codes 0 and 1 for exposure/developing; codes 0 and 1 for furcation visibility) could be determined for 110 teeth allocated to Control restorations (85%) and 81 Hall PMC teeth (63%). The reduction in visibility of the furcation area for Hall PMC teeth compared with the Control restoration teeth can be explained as the PMCs are fitted with no occlusal reduction, and so lengthen crown height, reducing the area of film adjacent to the furcation area.

There were 15 recorded episodes of pain in 13 children, related to study teeth. Twelve children reported pain from a tooth that had a Control restoration placed, with one of the children having two episodes of pain. Two children experienced pain from the Hall PMC tooth, with one of these children having pain from both the Control and the Hall PMC tooth (Figure 16). McNemar's test showed a significant difference in pain episodes experienced by a child for Control and Hall restorations (χ² = 7.692; P = 0.0055) with Hall PMCs giving less pain. The relative risk ratio of a Hall tooth giving pain was 0.15 compared with a conventionally restored tooth. This gave a numbers-needed-to-treat value of 12 (95% CI 7–33).

There were 57 Control restoration failures ascribed as 'Minor' (mean time to first failure -17.2 months; range – five to 35 months; median – 15 months) and six Hall PMCs (mean time to first failure – 18 months; range – six to 29 months; median – 18 months). Figure 17 summarises the 'Minor' failure data, and the types of failures for both Control restorations and Hall PMCs are shown in Table 13. McNemar's test demonstrated a statistically significant difference in 'Minor' failures (χ² = 45.45; P = 0.000) in favour of the intervention (Hall PMCs) allowing the null hypothesis of no difference in longevity of the Control restorations and Hall PMCs to be rejected. The relative risk ratio of a tooth with a Hall PMC suffering a 'Minor' failure was 0.09 compared to a conventionally restored tooth. This gave a numbers-needed-to-treat value of three (95% CI 2 to 3).

Table 14 details the numbers of Class I and Class II restorations and the materials used, where these were known and fitted the follow-up criteria. There was no statistically significant relationship between the use of glass ionomer for restoration of a Class I lesion and the risk of a 'Minor' failure (χ² = 1.208, P = 0.272). However, there was a statistically significant relationship for the use of glass ionomer to restore Class II lesions and a 'Minor' failure (χ² = 5.598, P = 0.018).

Restorative care alone will not solve the UK childhood dental caries problem, but for it to be even a part of the solution, it must be effective both in terms of managing the disease, and in its acceptability to children, their carers and dentists. The evidence that restorative care of primary teeth can be effective has, in many cases, been obtained from clinical trials run in Secondary Care or specialist practice settings, and it may be unrealistic to expect these findings to be generalisable to the general dental practice setting 418. To support the generalisability of the outcomes from this clinical trial, the setting was general dental practice, where the vast majority of children's dentistry in the UK is carried out. Clinical trials can be run successfully in Primary Care 34, although the difficulties of doing so are widely acknowledged 353637. However, the very high two year patient follow-up of 94% and the commitment in data gathering shown by the GDPs involved in this study are evidence of the possibilities of Primary Care based research.

In this trial, the study co-ordinator (NI) was also a GDP at the time of patient recruitment, and this seemed to aid recruitment through motivation of other GDPs in the study 38. The wide variation in patient recruitment by individual GDPs, with only four of the 18 GDPs meeting or exceeding their recruitment targets, mirrors the finding of the pilot study 26. Whether this was because, as is often the case with a new technique, it suited some GDPs clinical practice more than others, or whether this was related to barriers in the general practice environment to participating in research, is uncertain.

Children can find restorative care of carious primary molars difficult to accept. It is encouraging, therefore, that in 128 cases (97%) it was possible to fit a PMC using the Hall Technique, although in 15% of cases the Hall PMC was judged not to be fully seated. There was only one patient where it was not possible to fit a Hall PMC because of the child's inability to accept treatment, and this child also declined the Control restoration. All four teeth where no Hall PMC could be fitted were maxillary first primary molars, and 17 of the 18 cases where the PMC fit was judged radiographically as unsatisfactory were also first primary molars. Dentists using the technique report that mesial migration of second primary molars following loss of the distal marginal ridge of first primary molars can ma