Study selection and characteristics
The initial search yielded 1173 records, 616 duplicates were removed, and 557 records remained. Of these 520 were excluded as they were either: descriptive or qualitative; included a population that was outside, or did not include, the criterion age range; maternal interventions; prevalence studies; or workforce related. The search of the grey literature yielded four articles for review of which none met the inclusion or exclusion criteria. Hand searching yielded one article. Thirty-seven eligible articles remained of which 9 met the inclusion or exclusion criteria for this review. Figure 1 demonstrates study selection.
Table 1 summarises the quality assessment, design, population, interventions, ecological approach and community engagement intensity and outcomes of each study.
Quality Assessment
Two of the nine studies in this review were given an EPHPP Global Rating of strong (37, 38); two as moderate (39, 40), and the remaining five as weak (41-45). Additional file 1 describes the quality assessment components of each study. All but one of the included studies (42) were rated either moderate or strong for selection bias. The EPHPP tool defines a study as STRONG for selection bias (score=1) where it is ‘very likely’ that study participants were representative of the target population AND that there is greater than 80% participation from that population (46). Scores for selection bias increase the less likely it is that participants are representative of the target population. . The majority of studies were rated moderate (37, 40-43) or strong (38, 39) for study design.
Study Design
Study designs included three repeat cross-sectional studies (40, 44, 45) with one of these nested in a prevalence study (44); four pre-post studies (37, 41-43) with one nested in a mixed methods study (37); one randomized controlled trial (RCT) (38), and one cluster RCT (39) (Table 1). Due to the relatively small number, and heterogeneity, of the studies, no meta-analysis was performed.
Study population
Studies were conducted in diverse countries and geographical areas. Two studies were conducted in Australian rural or remote communities (37, 40); two studies in Taiwanese rural communities (39, 43) ; two in remote Canadian First Nations communities (42, 45); one in an American Indian (rural) setting (41), one in rural Brazil (44); and one in urban African (Nigerian) (38) communities (Table 1).
The sample size of studies varied between 17 - 324 participants. Where gender was reported (n=5) there was an even proportion of male and female participants. Five studies reported on outcomes for adolescents (38-41, 44), with two of these studies designed specifically for the age range included in our criterion ie 10-19 years (38, 39) (Table 1). The remainder reported results at a population level and did not specify results for participants in this age range.
Interventions
Eight out of the nine studies described intervention strategies targeting the individual (37-39, 41-45); four of these included the family (37, 41, 43, 45) as the enablers of change in the oral health status of their child (Table 1). This is important in this context as family and peer group (such as in the school setting) are considered part of the child’s interpersonal environment, and can be enablers for strategies targeting the individual (35). The remaining study reported on an environmental intervention (40) which was the provision of a water reticulation system including fluoride. Schools were the most common setting for study interventions. Three interventions were delivered only in schools (38, 39, 42); three included schools as one of a number of intervention settings (37, 41, 45); and one implying that a school was the setting (43).
Five intervention strategies emerged from the review. These included: 1) Educational (n=7) (37-39, 42-45) which targeted behaviour and knowledge of children and/or parents; 2) Clinical (n=4) (37, 41, 42, 44) which included fluoride varnish or rinse and dental treatment; 3) Provision of incentives (n=2), one using cash (45) and the other using ‘prizes’ (no details provided) (42); 4) Employment of local Health Workers (n=2) (37, 44); and 5) Reticulated fluoridated water supply (n=1) (40). Five studies delivered more than one intervention strategy (37, 42-45).
Ecological approach and Community Engagement Intensity
Only one study (37) was given an EAS score of four; indicating it included at least two strategy types and was implemented in more than three settings. Furthermore, no study reported a strong level of CEI. These were concerning results given that the features of these measures are recommended for conduct of research in Indigenous communities. When CEI features were examined six studies reported community governance or engagement in the research however little information on the nature of this was provided. Five studies reported that ‘capacity building’ occurred however when this feature was further examined capacity building was largely the formation of partnerships (37, 41, 42, 44, 45), with no studies describing career development pathways for Indigenous staff, and only one reporting on the participation of the community in decision making (41). No studies reported providing feedback of results to the participating communities. The number of studies which reported on each key feature of CEI are presented in Figure 2. Additionally, details of key feature of CEI can be found in Additional File 2. It should be noted that not all studies may have reported details of community engagement despite this being a key component of study design with Indigenous communities.
Outcomes
Eight of the nine studies reported statistically significant improvements in at least one component of oral health (Table 1). The most frequently reported outcome (n=5) was change in dmft/DMFT or the number of decayed, missing or filled tooth surfaces (dmfs/DMFS) (37, 40-42, 44); with two of these studies finding significant improvements of between p<0.001 and p<005 (40, 42). Two studies reported significant changes in oral health knowledge and/or behaviour (39, 43). Two studies reported a decline in caries prevalence; however, this was not tested for statistical significance (39,43). Six studies reported more than one outcome (37, 38, 40-44). A number of other statistically significant outcomes were reported across three studies including: reduction in treatment hours required (p<0.001) (45); reduced levels of debris, calculus and oral hygiene scores following video education compared to verbal education (p<0.05) (38); decreased levels of unmet restorative needs and increased numbers of fissure sealants (p<0.01) (37).