We enrolled 193 (93.2%) of 207 primary care providers from 34 practices groups throughout the health system (). This group of providers included 49 family medicine practitioners, 88 internists, 5 internists/pediatricians, 41 pediatricians, 9 physician assistants, and 1 nurse practitioner. One pediatrician from the intervention group withdrew following enrollment. Seventeen practices (88 providers) were randomly assigned to the intervention arm, and 17 practices (105 providers) were assigned control. From these practices, we identified 3,003 patients who as of April 30, 2007 met the enrollment criteria. Of these 3,003 patients, 87 refused participation in the study; 126 had their ICS medication stopped (and not restarted) in the baseline period from May 1, 2007 to August 1, 2007, and 92 disenrolled from the heath plan prior to the start of the intervention. Of the remaining 2,698 patients, 1,335 were affiliated with intervention practices (and hence had their ICS adherence information available to the participating providers at these practices) and 1,363 were associated with the control practices.
Flow chart of patients, providers, and practice clusters randomized in the intervention to provide clinicians with inhaled corticosteroid adherence information on their patients with asthma
The baseline characteristics of these patients are shown in . Only patient age differed significantly between patients in both study arms; average age was 26.8 years and 28.8 years for intervention arm and control arm patients, respectively (P=0.001). Baseline ICS adherence for the two comparison groups was similar at 25.6% and 27.7% for intervention arm and control arm patients, respectively (P=0.210).
Baseline characteristics of patients with asthma in the intervention and control study arms
At study end, adherence was not significantly higher among patients in the intervention arm when compared with patients in the control arm (23.2% vs. 24.9%, respectively; P-value = 0.553) (). As required, this analysis accounted for both practice cluster and randomization stratum, but additionally adjusting for patient age, sex, and race-ethnicity did not impact these finding (data not shown). The ICC for the primary outcome was 0.063. Adherence was significantly higher among participants in both arms who had had a visit to a primary care provider in the first 9 months of the study period when compared with participants with no such visit (23.2% vs. 12.0%; P=0.001 among individuals in the intervention arm; 24.9% vs. 15.2%; P=0.001 among individuals in the control arm). However, within these strata (i.e., individuals with no primary care visit in the first 9 months of the study and individuals with ≥1 visit), adherence was not significantly different at study end between individuals in the intervention and control arms (P=0.379 and P=0.639, respectively). Even individuals who had their general adherence information viewed had similar adherence at study end when compared to patients in the control arm (25.1% vs. 23.3%; P=0.585). In contrast, individuals who had their detailed adherence information viewed by their primary care provider in the first 9 months had significantly higher adherence at study end when compared with patients in the intervention arm whose provider did not view adherence information (35.7% vs. 12.3%; P=0.002) and when compared with patients in the control group (35.7% vs. 23.3%; P=0.026).
Differences in inhaled corticosteroid adherence at study end between patients with asthma in the intervention and control study arms*
shows the difference in ICS adherence at study end depending on whether in the first 9 months of the study a participant in the intervention group had a primary care visit, had their general adherence information viewed, or had their detailed adherence information viewed. ICS adherence was highest for participants who had all three, and this group had significantly higher adherence when compared to individuals who had both a primary care visit and a provider who viewed their general adherence information alone (i.e., group 6 vs. group 4; P=0.017). Adjusting these comparisons for baseline adherence levels had little substantive effect on the P-values presented with the exception that the Group 6 vs. Group 3 comparison became statistically significant (P=0.049) and the Group 2 vs. Group 1 comparison was no longer significant (P=0.058) (data not shown).
Differences in inhaled corticosteroid adherence for patients in the intervention group by whether their primary care provider viewed adherence information and whether they had a primary care visit*
Among the patients in the intervention arm, 127 subjects had 177 asthma-related emergency rooms visits, 10 subjects had 10 asthma-related hospitalizations, and 307 subjects had 550 separate occasions in which they were treated with an oral steroid during the study period. Among the patients in the control arm, 111 subjects had 159 asthma-related emergency rooms visits, 11 subjects had 11 asthma-related hospitalizations, and 300 subjects had 522 separate occasions in which they were treated with an oral steroid. Therefore, when compared to patients in the control arm, intervention patients did not differ in the time to first asthma-related emergency room visit (adjusted hazard ratio [aHR] 1.22; 95% CI 0.83–1.78), asthma-related hospitalization (aHR 0.86; 95% CI 0.32–2.29), or oral steroid use (aHR 1.07; 95% CI 0.89–1.29) (). Likewise, intervention patients had a similar rate of asthma-related emergency room visits (adjusted relative rate [aRR] 1.12; 95% CI 0.74–1.69), asthma-related hospitalizations (aRR 0.87; 95% CI 0.33–2.29), and oral steroid use (aRR 1.11; 95% CI 0.92–1.34) when compared to patients in the control arm during the study period.
Differences in secondary outcomes between patients with asthma in the intervention and control study arms*
In both intervention and control groups patients combined, those who had stable (i.e., unchanged) or improved adherence during the course of study had significantly lower relative rates of asthma-related emergency room visits (aRR 0.73; 95% CI 0.55–0.98) and oral steroid use (aRR 0.77; 95% CI 0.63–0.93) when compared with those whose adherence fell during this time ().
Differences in outcomes between patients who had stable or improved inhaled corticosteroid adherence during the study period compared with those who did not, stratified by study arm assignment*
As a post-hoc analysis, we also examined whether there were differences in the change in adherence (i.e., the change in adherence between the baseline period and study end) between study arms. As previously noted in , individuals in both arms had similar baseline levels of ICS adherence. We found that in general, the change in adherence over the study period did not differ between patients in the intervention group and those in the control group (Table E1 in the online supplemental material). Adherence fell over the study period for patients in both arms, and it fell to a lesser extent in those who had a primary care visit in the first 9 months of the study when compared to those who had no such visit. In contrast, among individuals in the intervention group, participants whose primary care provider viewed their detailed adherence information in the first 9 months of the study had an overall increase in their adherence over the study period, and this change was significantly different when compared to those whose primary care provider did not view their adherence data (3.8% vs. −6.7%, respectively; P=0.039).
Among participants in the intervention arm, the only patients that experienced an overall increase in ICS adherence over the study period were those whose primary care provider had viewed their detailed adherence information (Table E2 in the online supplemental material). Individuals in Group 6 (i.e., those with a clinic visit and who had their detailed adherence information viewed) showed significantly greater improvement when compared with individuals who did not have adherence information viewed, regardless of the presence of a clinic visit (i.e., Group 6 vs. Group 1, P=0.037; Group 6 vs. Group 2, P=0.029). These comparisons were similar for individuals in Group 5 (i.e., those with no clinic visit but who had their detailed adherence information viewed) but were of borderline statistical significance (i.e., Group 5 vs. Group 1, P=0.075; Group 5 vs. Group 2, P=0.074). The consistency of these findings when compared to the analyses presented in again suggests that individuals experienced an overall improvement in their adherence when providers viewed their detailed adherence information.