Four hundred sixty children were assessed for eligibility (See ), 295 were eligible (97% of those ineligible were not enrolled in the targeted schools), and 290 were randomized. Attrition was primarily due to children transferring to a non-participating school (98%). There was no significant difference between treatment groups in attrition rate (p=0.26). After randomization, children who discontinued use of budesonide in accordance with their physician’s order continued providing daily data.
Flow of participants through the study
The demographic and asthma characteristics of children randomized are shown in . There were no significant differences between the supervised therapy and usual care groups in any of the demographic or asthma characteristics. Children had an average age of 11 years, were 57% male and primarily black race (91%). Seventy-nine percent of children had moderate persistent asthma yet only 14% had rescue medication at school prior to the study. During the baseline period, there was no difference between the two treatment groups in the percent of expected budesonide refills that were filled (Supervised therapy = 57.3%; Usual Care = 54.2%, p=0.40). Because medication was administered daily by the study staff to the children in the supervised therapy group, we induced a high level of adherence among these children. However, the average adherence over the entire study period in the usual care group was 38% (SD = 25%). Further, 78% of children in the usual care group were ≤50% adherent indicating that the distribution of adherence is both highly variable and highly skewed.
Demographic Characteristics of Randomized Participants according to treatment group.
The internet based data collection system allowed us to collect high quality data with very few missing reports.36
During the study period, there were a total of 26,417 daily reports expected from the teachers and children. These daily reports provided information on peak flow meter readings, asthma symptoms, and absences. Of the observed reports, 25,744 (97.5%) provided data regarding asthma control. There were no adverse events related to the study drug reported.
There were no differences in the likelihood of experiencing an EPAC between the baseline and follow-up period in the usual care group (p=0.94); however, among those in the supervised therapy group, the odds of experiencing an EPAC during the baseline period were 1.57 times the odds of experiencing an EPAC during the follow-up period (90% CI: 1.20, 2.06, p=0.006). shows the decrease in percent of children experiencing an EPAC from baseline to follow-up by treatment group and shows the percent of children experiencing an EPAC each month by treatment group. shows results of the univariate analysis by month. Higher rates of EPACs were reported for the supervised therapy group than the usual care in November (not statistically significant) and December (p=0.04) of the baseline period.
Change in percent of children experiencing an episode of poor asthma control from baseline to intervention by treatment group.
Figure 3 Exacerbations by month: supervised therapy1 and usual care2
Episodes of poor asthma control by month, according to treatment group.
GEE modeling revealed a marginally significant interaction between the intervention and time period (p=0.065) indicating that children in the supervised therapy group showed greater improvement in asthma control from baseline to follow-up than children in the usual care group. The odds of experiencing an EPAC among those in the supervised group during the baseline period were significantly higher (1.38; 90% CI: 1.03, 1.87) than the odds of those in the usual care group during the same period, while the odds of experiencing an EPAC during the follow-up period among those in the supervised group were lower (0.89; 90% CI: 0.64, 1.22) than those in the usual care group during the baseline period. Looking at the individual components of the EPAC definition, there were no significant differences in either the interaction between or the main effects for treatment or time period (tests of interaction: PFM readings, p=0.20; absence due to respiratory illness, p=0.62; rescue medication, p=0.11).
There was no observed relationship between age (p=0.13), gender (p=0.18) or race (p=0.24) and the frequency of EPACs, nor was the relationship between intervention and the frequency of EPACs modified by including any of these covariates in the model. There was a significant relationship between school system (p=0.08) and the frequency with which EPACs occurred; however, these differences did not modify the relationship observed between the intervention and the frequency of EPACs. There was also a significant relationship between treatment and the likelihood of EPAC among specific subgroups of smoke exposure at baseline and emergency room visits and hospitalizations ( and ). For those exposed to smoke in the household, the magnitude of the interaction increases, compared to when the entire cohort is included. The p-value for the interaction in this group is 0.0997, suggesting that within those exposed to smoke in the household there is a difference in the frequency of EPACs between the usual and supervised groups. However, among those children either not exposed to smoke at baseline, or those exposed outside of the household only, there is no evidence of a treatment effect. Among those children who reported, at baseline, no ER visits or overnight hospitalizations during the past 12 months, there was a large difference in the effect of the intervention (p for interaction=0.0057). However, among those reporting at least one ER visit or overnight hospitalization, no differences in treatment were seen, either by year (i.e., no interaction), or overall. In these secondary analyses, subgroups had small samples; thus we lacked power to detect important differences in these groups.
Change in percent of children experiencing an episode of poor asthma control from baseline to intervention by secondhand smoke exposure and treatment group.
Change in percent of children experiencing an episode of poor asthma control from baseline to intervention by baseline ED/hospitalizations visits and treatment group.
shows the baseline and post-intervention QOL scores for both treatment groups. Significant increases in QOL were seen in overall scores as well as on domains for both treatment groups. There were no significant differences between the groups.
Figure 6 Mean change in quality of life overall score and subscales by treatment group*.