While we could find no comparative studies directly estimating improvements in outcomes associated with the use of an assistant, the studies included in systematic reviews of sealant effectiveness offer a potentially rich source of relevant information. These studies have met established rules of study design, conduct and measurement for inclusion in final bodies of evidence. In addition, they usually provide a detailed description of the intervention (for example, the preparation and placement procedures) and outcomes, in addition to the study participants, the time period and the setting.

A multivariate analysis of the association between the outcome in these studies (sealant retention) and four-handed delivery, in addition to other preparation and placement procedures, can provide indirect evidence of possible benefits. In the absence of randomized controlled trials, a multivariate approach can control for the effects of potential confounders measured in the studies, as well as provide estimates of the unique contribution of each procedure (such as four-handed delivery). Because such approaches may not account for all confounders, however, findings provide only indirect evidence of possible benefit. Information about the contribution of selected aspects of the sealant delivery protocol is important for clinical and public health decision making.

The primary objective of this secondary data analysis was to determine whether evidence existed that sealant retention increased with four-handed placement, while controlling for other factors that could affect retention. We chose retention instead of effectiveness as the outcome of interest, because retention would be affected less by differences in caries risk among the sample populations of multiple studies. In addition, the effectiveness of resin-based sealants is highly associated with retention, because these sealants act by providing a physical barrier that prevents microorganisms and food particles from collecting in pits and fissures.¹¹

	METHODS

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Definitions. We defined four-handed delivery as the placement of sealants by a primary operator with a second person present to provide assistance. Similarly, we defined two-handed delivery as the placement of sealants by a single operator. We used World Bank designations to classify countries where the studies were conducted as "high" income or "not high" income (a combination of low income, lower middle income and upper middle income).¹²

Inclusion criteria. We searched Medline and the Cochrane Library for systematic reviews of sealant effectiveness that were published in English between 1990 and 2005. Four systematic reviews,^13–16 which included 36 unique studies, met these inclusion criteria.^17–52 One reviewer (S.K.G.) screened these studies, and she excluded 25^28–52 for the following reasons: the study was not published in English⁵²; the study design was not a prospective cohort or randomized controlled trial⁴⁶; the study did not apply second- or third-generation sealant material^{28,30,32–40,42,45,49,50}; subjects were not between 5 and 10 years of age ⁴⁸; the study contained insufficient information to estimate both the percentage of sealants that were retained fully on permanent first molars according to year since placement and the standard errors (SEs) for these estimates^29,41,47,51; mechanical preparation, such as enameloplasty or fissureotomy, was performed before sealant placement⁴⁴; or lost or fractured sealant material was repaired or reapplied.^31,43

Data abstraction. The same reviewer (S.K.G.) abstracted the studies meeting the inclusion criteria. The abstraction form included the following factors hypothesized to be associated with sealant retention:

– two- or four-handed delivery;

– years since placement (for example, one, two or three);

– tooth-surface cleaning method (toothbrush or handpiece);

– isolation by cotton rolls or a rubber dam;

– type of suction;

– use of acid-etching and/or a bonding agent;

– type of primary operator (dentist or nondentist);

– income level of the country (high or not high).

We included the last factor to explore the assumption that greater access to and utilization of dental services, as well as differences in dental systems in higher-income countries, would increase the detection of incipient caries in sealed teeth. We contacted the authors of the studies to verify information about the conduct of the study if adequate detail was not provided in published reports.

Quality assessment. Because we selected studies from published systematic reviews that had explicit quality criteria for inclusion, we did not reassess all aspects of individual study quality but did document two selected quality aspects: number of primary operators and whether operators received training before delivering sealants to study subjects. It is important to remember that, to our knowledge, there are no comparative studies of sealant outcomes for two-versus four-handed placement and, thus, some commonly used criteria to determine study quality such as random allocation would not necessarily apply.

Outcome measure and data adjustment. Our outcome measure was retention at each annual follow-up examination of sealants that were placed on occlusal surfaces of first permanent molars. We defined retention as the presence of a sealant that completely covered the pits and fissures of the tooth. We used the following formula to calculate the SE of the retention rate:

where "n" represents the number of teeth initially sealed.

Because teeth in the same subject may be correlated with each other, conducting the analysis at the tooth level may have underestimated SEs. If a study provided only site-level retention data (for example, examiners reported multiple sites on individual teeth, such as buccolingual pits and mesiodistal occlusal pits), we used the reported retention rate but calculated the SE using the reported number of teeth instead of tooth sites. This adjustment resulted in higher SEs for studies using tooth sites as the unit of analysis.

Analysis. We calculated the summary-weighted retention rate separately for the studies that used two-and four-handed delivery for each of the three years after sealant placement. We weighted the studies by the reciprocal of their squared SE. To determine whether it was reasonable to pool the studies to attain a summary estimate of retention according to the presence or absence of a second operator for each of the three years, we examined whether the confidence intervals on the forest plots⁵³ overlapped for studies using two-handed delivery and for those using four-handed delivery.

We used weighted linear regression models to examine the effect of four-handed delivery alone (model 1) and in the presence of other hypothesized factors (model 2) on sealant retention for each year since placement. All explanatory factors were represented in the regression model as dichotomous independent variables, where "1" indicates the presence of the factor and "0" indicates the absence of the factor. We excluded hypothesized factors that were present in only one study, because the variable might have reflected other unique aspects of a single study. We considered explanatory variables to be significant if the P value for the coefficient was less than or equal to .05.

Because we had several possible combinations of explanatory variables and a small sample of studies, we constructed a tree diagram to determine for which combinations of variables we had studies. We also compared the explanatory power of model 1 (that is, how much total variation was explained by the model as measured by the adjusted R²) with that of model 2. We also reran the regression without the weights to determine whether the results still held when we weighted all of the studies equally.

	RESULTS

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
We included 11 studies in the final body of evidence (Table 1). Eight studies used four-handed delivery (representing 1,189 children and 1,944 teeth), while three used two-handed delivery (representing 885 children and 1,000 teeth). In nine studies, the operator performed prophylaxis using a handpiece (with pumice or prophylaxis paste) before placing the sealant. In two studies, the operator cleaned the tooth surfaces with a toothbrush and toothpaste. In six studies, dentists were the primary operators. Seven studies were conducted in high-income countries. Most studies began between 1973 and 1995. Four of the seven studies conducted in high-income countries began between 1973 and 1976. Of the remaining three studies, two likely began in 1977. The four studies published in countries with not-high incomes began between 1975 and 1995.

Retention rates. Summary retention rates for one, two and three years after placement were 89.0 percent (range, 73.4 to 94.6 percent), 81.2 percent (range, 59.5 to 88.9 percent) and 73.9 percent (range, 60.1 to 87.5 percent), respectively. Retention appeared to vary significantly according to study for both two- and four-handed studies (Figure). Summary retention rates for studies using four-handed delivery—equaling 89.8 percent after one year, 83.0 percent after two years and 83.0 percent after three years—were higher than summary retention rates for studies using two-handed delivery (equaling 84.8 percent after one year, 72.4 percent after two years and 67.9 percent after three years) (data not shown). For the regression model that included four-handed delivery and the time since sealant placement as explanatory variables (model 1 in Table 2 (page 287); 28 observations), the adjusted R² was 42 percent and the coefficient for four-handed delivery approached significance (P = .055).

View larger version (64K): [in this window] [in a new window]   Figure. Forest plots showing sealant retention (95 percent confidence interval) in studies involving two- and four-handed delivery, for each year after placement.

	DISCUSSION

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
The findings of this multivariate analysis indicate that, in comparison with two-handed delivery, four-handed delivery increased sealant retention by about 9 percentage points. It is important to note that we identified this positive association only when the variation in other selected factors (that is, time since sealant placement, provider type and surface cleaning method) was controlled across the studies. In contrast, the simple sealant retention rates in an individual study reflect multiple factors, and, thus, retention rates of more than 90 percent at one year for sealants placed in a study with two-handed delivery²¹ or less than 80 percent in a study with four-handed delivery^20,25 can be expected.

The forest plots suggest that significant heterogeneity existed among studies even after we stratified them according to the presence of a second operator. This likely reflects the multiple factors that can affect retention and thus indicated that the multivariate analysis, which controlled for the effects of some of these factors, was the appropriate approach. The high R²—ranging from 69 to 81 percent—for the final regression models indicates that these models included important variables affecting sealant retention in this group of studies.

The findings for some of the other variables in the model also were consistent with the initial hypotheses. First, sealant retention decreased over time. Three years after placement, about 15 percent of the sealants were completely or partially lost. In addition, sealants were less likely to be retained over time in high-income countries. As described above, greater use of dental services in these countries may have increased the probability of detecting caries in sealed teeth.

Unexpected findings. Certain findings of our analysis were unexpected. We found that hand-piece prophylaxis was associated with a reduction in sealant retention of about 20 percentage points when compared with toothbrush prophylaxis. Of the nine studies in the regression analysis that reported the use of a handpiece prophylaxis, five used prophylaxis paste, three used pumice and one did not specify. It is possible that some prophylaxis pastes marketed in the 1970s and 1980s may have contained oils or other substances that interfered with bonding. In addition, prophylaxis paste, along with pumice, may have been difficult to remove completely from the enamel surface before etching. In 1998, a study comparing toothbrush prophylaxis (with no toothpaste) with handpiece prophylaxis (with prophylaxis paste) reported similar rates of sealant retention—all greater than 97 percent—after one year.⁵⁴

Another unexpected finding was the association between having a dentist as the primary operator and lower sealant retention rates. The prevalence of sealant placement in the United States through the early 1990s, however, was less than 20 percent. This suggests that many operators likely had limited experience with sealant materials and/or placement techniques. The studies in which dentists were the primary operators may have been less likely to provide training in sealant placement than the studies in which the primary operators were nondentists for two possible reasons.

First, the investigators may have assumed that training was unnecessary because dentists generally have exceptional familiarity with restorative materials and techniques; moreover, even as early as the 1970s and 1980s, they were increasingly using resin-based composite materials. During that time, however, placement of resin-based composite materials generally was limited to restorations on smooth surfaces (that is, Class III, IV and V) with prepared margins. In the absence of training, some of the dentist operators and auxiliaries may not have appreciated fully the meticulousness and attention to detail that are required for successful sealant placement on pit-and-fissure surfaces.

Second, the opportunity cost of training time, as measured by foregone wages, would be higher for dentists than for nondentists. We cannot test this hypothesis because only three of the studies in this analysis specifically described the use of training before sealant placement. In the one study in which the dentists were trained, the retention rate was high, ranging from 95 percent at one year to 80 percent at three years after a one-time placement of sealants.¹⁸

Study limitations. This study and its underlying methodology have limitations. First, our comparison of the subgroups was observational. In the absence of random assignment in studies that were designed to directly compare sealant placement outcomes according to two- and four-handed delivery, the association between retention and an explanatory variable might have been due to another omitted causal variable, commonly known as confounding. Confounding may have been mitigated, however, because we used a multivariate analysis that attempted to control for key factors that are relevant to sealant retention.

Second, we did not have studies for all of the possible combinations of study factors, and there were, at most, two studies for any combination of factors. However, although the findings cannot be considered to be definitive because of potential confounding and the limited number of studies, the R² value suggests that, for this group of studies, the factors included in the model had good predictive power.

Third, our findings may be subject to recall bias because we contacted authors to obtain additional information if adequate data were not included in their report. For example, only five of the 11 studies reported the main explanatory variable—number of operators—in the original report.

Finally, our search universe was limited to studies included in systematic reviews of sealant effectiveness, and only one reviewer screened these studies. For this exploratory analysis, we chose a less resource-intensive method to identify and screen potential studies. In the absence of published comparative studies, this approach is attractive because it provides an efficient method of collecting data from well-conducted studies. The studies included in systematic reviews have met rules of study design, conduct and measurement. In addition, we minimized bias in selecting studies for the current analysis, because the universe of studies was determined by authors of the original systematic reviews. Inclusion and exclusion criteria in this analysis were objective and were specified before we screened available studies. Findings may be useful in developing hypotheses and directing resources for further research.

	CONCLUSIONS

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
For this group of 11 studies, four-handed delivery was associated with higher retention of resin-based sealants. Although these descriptive findings cannot be generalized to all settings, they justify allocating resources to studies that directly compare sealant placement outcomes using two- and four-handed delivery.