Van de Hoef 2007.

Methods	Design: cluster, parallel RCT Number of participant: 299 Setting: not reported Country: Surinam Unit of randomisation: child Unit of analysis: tooth Follow‐up: 6 and 30 months Dropout: 51.7% after 30 months
Participants	Number randomised (participants): 299 children (153 ART group and 146 CT group)/408 teeth (205 ART and 203 CT) Number analysed: 211 teeth Age mean and SD (range): 7.5 years (6.0‐12.9 years) Gender: female 155 (51.8%), male 144 (48.2%) Average dmft score: not reported Dentition: primary Type of caries lesion: multiple surface caries lesion Inclusion criteria: schoolchildren in good mental and physical health with ≥ 1 small proximally situated cavity in a primary molar that was accessible to hand instruments from the occlusal surface and where no pulp exposure was expected. The measurements of the cavity had to be < 1 mm mesio‐distally and 2 mm in bucco‐lingual/palatinal direction. The antagonist tooth had to be present. Exclusion criteria: pain, swelling or fistula
Interventions	The study had four arms: Group 1: ART approach + H‐GIC Group 2: ART approach + H‐GIC with local anaesthesia Group 3; CT + H‐GIC with local anaesthesia. Group 4: CT + H‐GIC Children in the ART approach were treated using only hand instruments (i.e. hatchets and spoon excavators) to remove the caries lesions. Participants in the CT group were treated with rotary instruments, i.e. stainless steel round burs in a slow handpiece without water cooling. After access to the cavity was obtained, at first the enamel‐dentine border was cleaned and after that the remaining caries was removed. In both treatments after finishing the preparation a piece of metal matrix band (Matricodent) was applied and fixed with a wooden wedge. In all cases hand‐mixed glass ionomer (Fuji IX, GC Corporation) was used as restoration material. The interventions were conducted by one dentist, one dental student and two hygienists.
Outcomes	Success was evaluated through ART criteria after 6 and 30 months Discomfort assessed by modified Venham scale and heart frequency at seven fixed moments during dental treatment: (i) during entrance in the treatment room, (ii) during local analgesia (in groups 2 and 4), (iii) at the start of preparation, (iv) during deep excavation, (v) during application of the matrix and wedge, (vi) at the start of restoration (when glass ionomer was applied), (vii) at the end of restoration
Notes	Funding: Foundation of Youth Dental Care in Paramaribo, Suriname and GC company provided the GIC Trial register number not reported Samples size not calculated Intraexaminer consistency values range from 0.73‐0.84 (Cohen’s kappa) Interexaminer consistency was calculated: 0.72 for the 6‐month evaluation and 0.93 for the evaluation after 30 months. Some of the children received a second restoration placed in another molar. In these cases the same treatment protocol for both restorations was used.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The children were randomly divided into four treatment groups" "The randomization list was obtained by means of SPSS"
Allocation concealment (selection bias)	Unclear risk	Comments: not reported
Blinding of participants and personnel (performance bias) ‐ participant	High risk	Comments: no information provided, but the participants could tell whether manual or rotary instruments were used
Blinding of participants and personnel (performance bias) ‐ operator All outcomes	High risk	Comments: blinding not possible ‐ operator knew the intervention
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The restorations were evaluated by two final‐year dental students of ACTA (who did not perform any treatment)"
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote: "The majority of the dropouts concerned absent patients and shed teeth" Comments: loss to follow‐up close to 50% at 30 months. How many losses due to absence or shedding not reported
Selective reporting (reporting bias)	High risk	Comments: discomfort was not reported at all measured times, only during deep excavation and restoration. Not was included a mean of all measured.
Other bias	High risk	Comments: baseline characteristics or details about co‐interventions not reported. The analysis did not consider the intra‐cluster correlation coefficient.