Resumo Pouco se sabe sobre o impacto de um delineamento cruzado nos desfechos de ensaios clínicos randomizados voltados ao comportamento de crianças durante tratamento odontológico. Este estudo objetivou avaliar o efeito da sequência de administração do sedativo, da droga em si e da idade dos participantes no comportamento de crianças que receberam duas consultas odontológicas consecutivas. Dezoito crianças saudáveis não colaboradoras, 2-5 anos de idade, foram randomizadas em dois grupos: G1 - 1,0 mg/kg midazolam oral (primeira sessão) e placebo oral (segunda sessão); G2 - placebo (primeira) e 1,0 mg/kg midazolam oral (segunda). Um observador treinado avaliou o comportamento infantil. Os dados foram analisados por ANOVA de três fatores (alfa=0,05). Midazolam [média(DP); 71,7%(16,5)] e placebo [48,6%(33,1)] resultaram em mais comportamento não cooperativo quando administrados na primeira sessão comparado com a segunda (p=0,001). Com o uso do placebo, crianças de 2-3 anos de idade exibiram mais comportamento não cooperativo [G1 54,9%(36,2); G2 80,5%(8,3)] que as de 4-5 anos de idade (p=0,04). Além disso, a porcentagem de redução do comportamento não cooperativo foi maior em crianças mais velhas em G1 (76,2%) e em crianças mais novas em G2 (32,9%). Considerando a avaliação do comportamento infantil sob sedação, a primeira sessão odontológica influenciou a segunda visita. Os resultados deste estudo confirmam a especulação de que o delineamento cruzado é inadequado para avaliar o comportamento odontológico relacionado à ansiedade/comportamento infantil; os resultados dos ensaios cruzados de sedação odontológica devem ser tratados com extrema cautela.

Abstract Crossover studies continue to be published in spite of warnings about their inherent risks in relation to behavioral outcomes. This study took the opportunity of access to secondary data analysis in order to demonstrate the impact of a crossover design on the outcomes of randomized clinical trials aimed at the behavior of children during dental treatment. We evaluated the effect of the sequence of sedative administration, the sedative and the participant’s age on the behavior of children undergoing two sequential dental visits. Eighteen uncooperative healthy young children were equally randomly assigned to: (G1) 1.0 mg/kg oral midazolam (first session) and oral placebo (second session); (G2) oral placebo (first) and 1.0 mg/kg oral midazolam (second). One trained observer assessed children’s behavior. Data were analyzed by three-way mixed ANOVA. Both midazolam [mean(SD); 71.7%(16.5)] and placebo [48.6%(33.1)] produced more struggling behavior when they were administered in the first session compared to the second one (p=0.001). For the placebo, children aged 2-3 years exhibited more struggling behavior [G1 54.9%(36.2); G2 80.5%(8.3)] than those aged 4-5 years (p=0.04). Also, the reduction of percentage of struggling behavior was higher in G1 for older children (76.2%) and in G2 for younger children (32.9%). There were significant interactions between drug and sequence of administration, and between drug and age. The results of our study confirm the conventional wisdom that crossover study design is inappropriate to evaluate children’s behavior/anxiety related-dental treatment under sedation and the results of crossover studies of dental sedation should be treated with extreme caution.

Little is known about whether midazolam sedation can reduce salivary cortisol levels and consequently influence children’s behaviour during dental treatment. The aim of this study was to evaluate the effect of midazolam sedation on salivary cortisol and its correlation with children’s behaviour during restorative dental treatment. Eighteen healthy children, aged two to five years, were randomly assigned to two dental treatment appointments, both with physical restraint: oral midazolam 1 mg/kg (MS) and placebo (PS). An observer assessed the children’s behaviour (videos) using the Ohio State University Behavioral Rating Scale (OSUBRS). The children’s saliva was collected just after waking up, on arrival at the dental school, 25 minutes after local anaesthesia, and 25 minutes after the end of the procedure. Salivary cortisol levels were determined using the enzyme-linked immunoabsorbent assay. The data were analysed by bivariate tests and multivariate analysis of variance (5% level). Salivary cortisol levels were lower in the MS group than in the PS group at the time of anaesthesia (p = 0.004), but did not vary during the appointment within sedation (p = 0.319) or placebo (p = 0.080) groups. Children’s behaviour was negative most of the time and did not differ between MS and PS; however, the behaviour (OSUBRS) did not correlate with salivary cortisol levels. Oral midazolam is able to control salivary cortisol levels during dental treatment of pre-schoolers, which might not lead to better clinical behaviour.